github/tinygrad
1
1
Fork 0
mirror of https://github.com/tinygrad/tinygrad.git synced 2026-01-09 06:58:11 -05:00
Code Issues Packages Projects Releases Wiki Activity
Files
05d27abcc2be5706130d3f9dec31b62f5bc48361
tinygrad/extra/assembly/amd/autogen/rdna3/gen_pcode.py
George Hotz 05d27abcc2 tests pass
2025-12-30 13:49:05 +00:00

10122 lines
377 KiB
Python
Raw Blame History

# autogenerated by pcode.py - do not edit
# to regenerate: python -m extra.assembly.amd.pcode --arch rdna3
# ruff: noqa: E501,F405,F403
# mypy: ignore-errors
from extra.assembly.amd.autogen.rdna3 import SOP1Op, SOP2Op, SOPCOp, SOPKOp, SOPPOp, VOP1Op, VOP2Op, VOP3Op, VOP3SDOp, VOP3POp, VOPCOp
from extra.assembly.amd.pcode import *
def _SOP1Op_S_MOV_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.b32 = S0.b32
# --- compiled pseudocode ---
D0.b32 = S0.b32
# --- end pseudocode ---
def _SOP1Op_S_MOV_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.b64 = S0.b64
# --- compiled pseudocode ---
D0.b64 = S0.b64
# --- end pseudocode ---
def _SOP1Op_S_CMOV_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if SCC then
# D0.b32 = S0.b32
# endif
# --- compiled pseudocode ---
if SCC:
D0.b32 = S0.b32
# --- end pseudocode ---
def _SOP1Op_S_CMOV_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if SCC then
# D0.b64 = S0.b64
# endif
# --- compiled pseudocode ---
if SCC:
D0.b64 = S0.b64
# --- end pseudocode ---
def _SOP1Op_S_BREV_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32[31 : 0] = S0.u32[0 : 31]
# --- compiled pseudocode ---
D0.u32[31 : 0] = S0.u32[0 : 31]
# --- end pseudocode ---
def _SOP1Op_S_BREV_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[63 : 0] = S0.u64[0 : 63]
# --- compiled pseudocode ---
D0.u64[63 : 0] = S0.u64[0 : 63]
# --- end pseudocode ---
def _SOP1Op_S_CTZ_I32_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = -1;
# // Set if no ones are found
# for i in 0 : 31 do
# // Search from LSB
# if S0.u32[i] == 1'1U then
# tmp = i;
# endif
# endfor;
# D0.i32 = tmp
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(-1)
for i in range(0, int(31)+1):
if S0.u32[i] == 1:
tmp._val = i; break
D0.i32 = tmp
# --- end pseudocode ---
def _SOP1Op_S_CTZ_I32_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = -1;
# // Set if no ones are found
# for i in 0 : 63 do
# // Search from LSB
# if S0.u64[i] == 1'1U then
# tmp = i;
# endif
# endfor;
# D0.i32 = tmp
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(-1)
for i in range(0, int(63)+1):
if S0.u64[i] == 1:
tmp._val = i; break
D0.i32 = tmp
# --- end pseudocode ---
def _SOP1Op_S_CLZ_I32_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = -1;
# // Set if no ones are found
# for i in 0 : 31 do
# // Search from MSB
# if S0.u32[31 - i] == 1'1U then
# tmp = i;
# endif
# endfor;
# D0.i32 = tmp
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(-1)
for i in range(0, int(31)+1):
if S0.u32[31 - i] == 1:
tmp._val = i; break
D0.i32 = tmp
# --- end pseudocode ---
def _SOP1Op_S_CLZ_I32_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = -1;
# // Set if no ones are found
# for i in 0 : 63 do
# // Search from MSB
# if S0.u64[63 - i] == 1'1U then
# tmp = i;
# endif
# endfor;
# D0.i32 = tmp
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(-1)
for i in range(0, int(63)+1):
if S0.u64[63 - i] == 1:
tmp._val = i; break
D0.i32 = tmp
# --- end pseudocode ---
def _SOP1Op_S_CLS_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = -1;
# // Set if all bits are the same
# for i in 1 : 31 do
# // Search from MSB
# if S0.u32[31 - i] != S0.u32[31] then
# tmp = i;
# endif
# endfor;
# D0.i32 = tmp
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(-1)
for i in range(1, int(31)+1):
if S0.u32[31 - i] != S0.u32[31]:
tmp = Reg(i)
D0.i32 = tmp
# --- end pseudocode ---
def _SOP1Op_S_CLS_I32_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = -1;
# // Set if all bits are the same
# for i in 1 : 63 do
# // Search from MSB
# if S0.u64[63 - i] != S0.u64[63] then
# tmp = i;
# endif
# endfor;
# D0.i32 = tmp
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(-1)
for i in range(1, int(63)+1):
if S0.u64[63 - i] != S0.u64[63]:
tmp = Reg(i)
D0.i32 = tmp
# --- end pseudocode ---
def _SOP1Op_S_SEXT_I32_I8(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = 32'I(signext(S0.i8))
# --- compiled pseudocode ---
D0.i32 = (signext(S0.i8))
# --- end pseudocode ---
def _SOP1Op_S_SEXT_I32_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = 32'I(signext(S0.i16))
# --- compiled pseudocode ---
D0.i32 = (signext(S0.i16))
# --- end pseudocode ---
def _SOP1Op_S_BITSET0_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32[S0.u32[4 : 0]] = 1'0U
# --- compiled pseudocode ---
D0.u32[S0.u32[4 : 0]] = 0
# --- end pseudocode ---
def _SOP1Op_S_BITSET0_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[S0.u32[5 : 0]] = 1'0U
# --- compiled pseudocode ---
D0.u64[S0.u32[5 : 0]] = 0
# --- end pseudocode ---
def _SOP1Op_S_BITSET1_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32[S0.u32[4 : 0]] = 1'1U
# --- compiled pseudocode ---
D0.u32[S0.u32[4 : 0]] = 1
# --- end pseudocode ---
def _SOP1Op_S_BITSET1_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[S0.u32[5 : 0]] = 1'1U
# --- compiled pseudocode ---
D0.u64[S0.u32[5 : 0]] = 1
# --- end pseudocode ---
def _SOP1Op_S_BITREPLICATE_B64_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = S0.u32;
# for i in 0 : 31 do
# D0.u64[i * 2] = tmp[i];
# D0.u64[i * 2 + 1] = tmp[i]
# endfor
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(S0.u32)
for i in range(0, int(31)+1):
D0.u64[i * 2] = tmp[i]
D0.u64[i * 2 + 1] = tmp[i]
# --- end pseudocode ---
def _SOP1Op_S_ABS_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = S0.i32 < 0 ? -S0.i32 : S0.i32;
# SCC = D0.i32 != 0
# --- compiled pseudocode ---
D0.i32 = ((-S0.i32) if (S0.i32 < 0) else (S0.i32))
SCC.b32 = D0.i32 != 0
# --- end pseudocode ---
def _SOP1Op_S_BCNT0_I32_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = 0;
# for i in 0 : 31 do
# tmp += S0.u32[i] == 1'0U ? 1 : 0
# endfor;
# D0.i32 = tmp;
# SCC = D0.u32 != 0U
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(0)
for i in range(0, int(31)+1):
tmp += ((1) if (S0.u32[i] == 0) else (0))
D0.i32 = tmp
SCC.b32 = D0.u32 != 0
# --- end pseudocode ---
def _SOP1Op_S_BCNT0_I32_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = 0;
# for i in 0 : 63 do
# tmp += S0.u64[i] == 1'0U ? 1 : 0
# endfor;
# D0.i32 = tmp;
# SCC = D0.u64 != 0ULL
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(0)
for i in range(0, int(63)+1):
tmp += ((1) if (S0.u64[i] == 0) else (0))
D0.i32 = tmp
SCC.b32 = D0.u64 != 0
# --- end pseudocode ---
def _SOP1Op_S_BCNT1_I32_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = 0;
# for i in 0 : 31 do
# tmp += S0.u32[i] == 1'1U ? 1 : 0
# endfor;
# D0.i32 = tmp;
# SCC = D0.u32 != 0U
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(0)
for i in range(0, int(31)+1):
tmp += ((1) if (S0.u32[i] == 1) else (0))
D0.i32 = tmp
SCC.b32 = D0.u32 != 0
# --- end pseudocode ---
def _SOP1Op_S_BCNT1_I32_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = 0;
# for i in 0 : 63 do
# tmp += S0.u64[i] == 1'1U ? 1 : 0
# endfor;
# D0.i32 = tmp;
# SCC = D0.u64 != 0ULL
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(0)
for i in range(0, int(63)+1):
tmp += ((1) if (S0.u64[i] == 1) else (0))
D0.i32 = tmp
SCC.b32 = D0.u64 != 0
# --- end pseudocode ---
def _SOP1Op_S_QUADMASK_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = 0U;
# for i in 0 : 7 do
# tmp[i] = S0.u32[i * 4 +: 4] != 0U
# endfor;
# D0.u32 = tmp;
# SCC = D0.u32 != 0U
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(0)
for i in range(0, int(7)+1):
tmp[i] = S0.u32[(i * 4) + (4) - 1 : (i * 4)] != 0
D0.u32 = tmp
SCC.b32 = D0.u32 != 0
# --- end pseudocode ---
def _SOP1Op_S_QUADMASK_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = 0ULL;
# for i in 0 : 15 do
# tmp[i] = S0.u64[i * 4 +: 4] != 0ULL
# endfor;
# D0.u64 = tmp;
# SCC = D0.u64 != 0ULL
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(0)
for i in range(0, int(15)+1):
tmp[i] = S0.u64[(i * 4) + (4) - 1 : (i * 4)] != 0
D0.u64 = tmp
SCC.b32 = D0.u64 != 0
# --- end pseudocode ---
def _SOP1Op_S_WQM_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = 0U;
# declare i : 6'U;
# for i in 6'0U : 6'31U do
# tmp[i] = S0.u32[i & 6'60U +: 6'4U] != 0U
# endfor;
# D0.u32 = tmp;
# SCC = D0.u32 != 0U
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(0)
for i in range(0, int(31)+1):
tmp[i] = S0.u32[(i & 60) + (4) - 1 : (i & 60)] != 0
D0.u32 = tmp
SCC.b32 = D0.u32 != 0
# --- end pseudocode ---
def _SOP1Op_S_WQM_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = 0ULL;
# declare i : 6'U;
# for i in 6'0U : 6'63U do
# tmp[i] = S0.u64[i & 6'60U +: 6'4U] != 0ULL
# endfor;
# D0.u64 = tmp;
# SCC = D0.u64 != 0ULL
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(0)
for i in range(0, int(63)+1):
tmp[i] = S0.u64[(i & 60) + (4) - 1 : (i & 60)] != 0
D0.u64 = tmp
SCC.b32 = D0.u64 != 0
# --- end pseudocode ---
def _SOP1Op_S_NOT_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = ~S0.u32;
# SCC = D0.u32 != 0U
# --- compiled pseudocode ---
D0.u32 = ~S0.u32
SCC.b32 = D0.u32 != 0
# --- end pseudocode ---
def _SOP1Op_S_NOT_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64 = ~S0.u64;
# SCC = D0.u64 != 0ULL
# --- compiled pseudocode ---
D0.u64 = ~S0.u64
SCC.b32 = D0.u64 != 0
# --- end pseudocode ---
def _SOP1Op_S_AND_SAVEEXEC_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Calculate bitwise AND on the scalar input and the EXEC mask, store the calculated result into the EXEC mask,
# set SCC iff the calculated result is nonzero and store the original value of the EXEC mask into the scalar
# saveexec = EXEC.u32;
# EXEC.u32 = (S0.u32 & EXEC.u32);
# D0.u32 = saveexec.u32;
# SCC = EXEC.u32 != 0U
saveexec = Reg(EXEC._val)
# --- compiled pseudocode ---
saveexec = Reg(EXEC.u32)
EXEC.u32 = (S0.u32 & EXEC.u32)
D0.u32 = saveexec.u32
SCC.b32 = EXEC.u32 != 0
# --- end pseudocode ---
def _SOP1Op_S_AND_SAVEEXEC_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Calculate bitwise AND on the scalar input and the EXEC mask, store the calculated result into the EXEC mask,
# set SCC iff the calculated result is nonzero and store the original value of the EXEC mask into the scalar
# saveexec = EXEC.u64;
# EXEC.u64 = (S0.u64 & EXEC.u64);
# D0.u64 = saveexec.u64;
# SCC = EXEC.u64 != 0ULL
saveexec = Reg(EXEC._val)
# --- compiled pseudocode ---
saveexec = Reg(EXEC.u64)
EXEC.u64 = (S0.u64 & EXEC.u64)
D0.u64 = saveexec.u64
SCC.b32 = EXEC.u64 != 0
# --- end pseudocode ---
def _SOP1Op_S_OR_SAVEEXEC_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Calculate bitwise OR on the scalar input and the EXEC mask, store the calculated result into the EXEC mask, set
# SCC iff the calculated result is nonzero and store the original value of the EXEC mask into the scalar destination
# saveexec = EXEC.u32;
# EXEC.u32 = (S0.u32 | EXEC.u32);
# D0.u32 = saveexec.u32;
# SCC = EXEC.u32 != 0U
saveexec = Reg(EXEC._val)
# --- compiled pseudocode ---
saveexec = Reg(EXEC.u32)
EXEC.u32 = (S0.u32 | EXEC.u32)
D0.u32 = saveexec.u32
SCC.b32 = EXEC.u32 != 0
# --- end pseudocode ---
def _SOP1Op_S_OR_SAVEEXEC_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Calculate bitwise OR on the scalar input and the EXEC mask, store the calculated result into the EXEC mask, set
# SCC iff the calculated result is nonzero and store the original value of the EXEC mask into the scalar destination
# saveexec = EXEC.u64;
# EXEC.u64 = (S0.u64 | EXEC.u64);
# D0.u64 = saveexec.u64;
# SCC = EXEC.u64 != 0ULL
saveexec = Reg(EXEC._val)
# --- compiled pseudocode ---
saveexec = Reg(EXEC.u64)
EXEC.u64 = (S0.u64 | EXEC.u64)
D0.u64 = saveexec.u64
SCC.b32 = EXEC.u64 != 0
# --- end pseudocode ---
def _SOP1Op_S_XOR_SAVEEXEC_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Calculate bitwise XOR on the scalar input and the EXEC mask, store the calculated result into the EXEC mask,
# set SCC iff the calculated result is nonzero and store the original value of the EXEC mask into the scalar
# saveexec = EXEC.u32;
# EXEC.u32 = (S0.u32 ^ EXEC.u32);
# D0.u32 = saveexec.u32;
# SCC = EXEC.u32 != 0U
saveexec = Reg(EXEC._val)
# --- compiled pseudocode ---
saveexec = Reg(EXEC.u32)
EXEC.u32 = (S0.u32 ^ EXEC.u32)
D0.u32 = saveexec.u32
SCC.b32 = EXEC.u32 != 0
# --- end pseudocode ---
def _SOP1Op_S_XOR_SAVEEXEC_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Calculate bitwise XOR on the scalar input and the EXEC mask, store the calculated result into the EXEC mask,
# set SCC iff the calculated result is nonzero and store the original value of the EXEC mask into the scalar
# saveexec = EXEC.u64;
# EXEC.u64 = (S0.u64 ^ EXEC.u64);
# D0.u64 = saveexec.u64;
# SCC = EXEC.u64 != 0ULL
saveexec = Reg(EXEC._val)
# --- compiled pseudocode ---
saveexec = Reg(EXEC.u64)
EXEC.u64 = (S0.u64 ^ EXEC.u64)
D0.u64 = saveexec.u64
SCC.b32 = EXEC.u64 != 0
# --- end pseudocode ---
def _SOP1Op_S_NAND_SAVEEXEC_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Calculate bitwise NAND on the scalar input and the EXEC mask, store the calculated result into the EXEC mask,
# set SCC iff the calculated result is nonzero and store the original value of the EXEC mask into the scalar
# saveexec = EXEC.u32;
# EXEC.u32 = ~(S0.u32 & EXEC.u32);
# D0.u32 = saveexec.u32;
# SCC = EXEC.u32 != 0U
saveexec = Reg(EXEC._val)
# --- compiled pseudocode ---
saveexec = Reg(EXEC.u32)
EXEC.u32 = ~(S0.u32 & EXEC.u32)
D0.u32 = saveexec.u32
SCC.b32 = EXEC.u32 != 0
# --- end pseudocode ---
def _SOP1Op_S_NAND_SAVEEXEC_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Calculate bitwise NAND on the scalar input and the EXEC mask, store the calculated result into the EXEC mask,
# set SCC iff the calculated result is nonzero and store the original value of the EXEC mask into the scalar
# saveexec = EXEC.u64;
# EXEC.u64 = ~(S0.u64 & EXEC.u64);
# D0.u64 = saveexec.u64;
# SCC = EXEC.u64 != 0ULL
saveexec = Reg(EXEC._val)
# --- compiled pseudocode ---
saveexec = Reg(EXEC.u64)
EXEC.u64 = ~(S0.u64 & EXEC.u64)
D0.u64 = saveexec.u64
SCC.b32 = EXEC.u64 != 0
# --- end pseudocode ---
def _SOP1Op_S_NOR_SAVEEXEC_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Calculate bitwise NOR on the scalar input and the EXEC mask, store the calculated result into the EXEC mask,
# set SCC iff the calculated result is nonzero and store the original value of the EXEC mask into the scalar
# saveexec = EXEC.u32;
# EXEC.u32 = ~(S0.u32 | EXEC.u32);
# D0.u32 = saveexec.u32;
# SCC = EXEC.u32 != 0U
saveexec = Reg(EXEC._val)
# --- compiled pseudocode ---
saveexec = Reg(EXEC.u32)
EXEC.u32 = ~(S0.u32 | EXEC.u32)
D0.u32 = saveexec.u32
SCC.b32 = EXEC.u32 != 0
# --- end pseudocode ---
def _SOP1Op_S_NOR_SAVEEXEC_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Calculate bitwise NOR on the scalar input and the EXEC mask, store the calculated result into the EXEC mask,
# set SCC iff the calculated result is nonzero and store the original value of the EXEC mask into the scalar
# saveexec = EXEC.u64;
# EXEC.u64 = ~(S0.u64 | EXEC.u64);
# D0.u64 = saveexec.u64;
# SCC = EXEC.u64 != 0ULL
saveexec = Reg(EXEC._val)
# --- compiled pseudocode ---
saveexec = Reg(EXEC.u64)
EXEC.u64 = ~(S0.u64 | EXEC.u64)
D0.u64 = saveexec.u64
SCC.b32 = EXEC.u64 != 0
# --- end pseudocode ---
def _SOP1Op_S_XNOR_SAVEEXEC_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Calculate bitwise XNOR on the scalar input and the EXEC mask, store the calculated result into the EXEC mask,
# set SCC iff the calculated result is nonzero and store the original value of the EXEC mask into the scalar
# saveexec = EXEC.u32;
# EXEC.u32 = ~(S0.u32 ^ EXEC.u32);
# D0.u32 = saveexec.u32;
# SCC = EXEC.u32 != 0U
saveexec = Reg(EXEC._val)
# --- compiled pseudocode ---
saveexec = Reg(EXEC.u32)
EXEC.u32 = ~(S0.u32 ^ EXEC.u32)
D0.u32 = saveexec.u32
SCC.b32 = EXEC.u32 != 0
# --- end pseudocode ---
def _SOP1Op_S_XNOR_SAVEEXEC_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Calculate bitwise XNOR on the scalar input and the EXEC mask, store the calculated result into the EXEC mask,
# set SCC iff the calculated result is nonzero and store the original value of the EXEC mask into the scalar
# saveexec = EXEC.u64;
# EXEC.u64 = ~(S0.u64 ^ EXEC.u64);
# D0.u64 = saveexec.u64;
# SCC = EXEC.u64 != 0ULL
saveexec = Reg(EXEC._val)
# --- compiled pseudocode ---
saveexec = Reg(EXEC.u64)
EXEC.u64 = ~(S0.u64 ^ EXEC.u64)
D0.u64 = saveexec.u64
SCC.b32 = EXEC.u64 != 0
# --- end pseudocode ---
def _SOP1Op_S_AND_NOT0_SAVEEXEC_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Calculate bitwise AND on the EXEC mask and the negation of the scalar input, store the calculated result into
# the EXEC mask, set SCC iff the calculated result is nonzero and store the original value of the EXEC mask into
# saveexec = EXEC.u32;
# EXEC.u32 = (~S0.u32 & EXEC.u32);
# D0.u32 = saveexec.u32;
# SCC = EXEC.u32 != 0U
saveexec = Reg(EXEC._val)
# --- compiled pseudocode ---
saveexec = Reg(EXEC.u32)
EXEC.u32 = (~S0.u32 & EXEC.u32)
D0.u32 = saveexec.u32
SCC.b32 = EXEC.u32 != 0
# --- end pseudocode ---
def _SOP1Op_S_AND_NOT0_SAVEEXEC_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Calculate bitwise AND on the EXEC mask and the negation of the scalar input, store the calculated result into
# the EXEC mask, set SCC iff the calculated result is nonzero and store the original value of the EXEC mask into
# saveexec = EXEC.u64;
# EXEC.u64 = (~S0.u64 & EXEC.u64);
# D0.u64 = saveexec.u64;
# SCC = EXEC.u64 != 0ULL
saveexec = Reg(EXEC._val)
# --- compiled pseudocode ---
saveexec = Reg(EXEC.u64)
EXEC.u64 = (~S0.u64 & EXEC.u64)
D0.u64 = saveexec.u64
SCC.b32 = EXEC.u64 != 0
# --- end pseudocode ---
def _SOP1Op_S_OR_NOT0_SAVEEXEC_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Calculate bitwise OR on the EXEC mask and the negation of the scalar input, store the calculated result into the
# EXEC mask, set SCC iff the calculated result is nonzero and store the original value of the EXEC mask into the
# saveexec = EXEC.u32;
# EXEC.u32 = (~S0.u32 | EXEC.u32);
# D0.u32 = saveexec.u32;
# SCC = EXEC.u32 != 0U
saveexec = Reg(EXEC._val)
# --- compiled pseudocode ---
saveexec = Reg(EXEC.u32)
EXEC.u32 = (~S0.u32 | EXEC.u32)
D0.u32 = saveexec.u32
SCC.b32 = EXEC.u32 != 0
# --- end pseudocode ---
def _SOP1Op_S_OR_NOT0_SAVEEXEC_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Calculate bitwise OR on the EXEC mask and the negation of the scalar input, store the calculated result into the
# EXEC mask, set SCC iff the calculated result is nonzero and store the original value of the EXEC mask into the
# saveexec = EXEC.u64;
# EXEC.u64 = (~S0.u64 | EXEC.u64);
# D0.u64 = saveexec.u64;
# SCC = EXEC.u64 != 0ULL
saveexec = Reg(EXEC._val)
# --- compiled pseudocode ---
saveexec = Reg(EXEC.u64)
EXEC.u64 = (~S0.u64 | EXEC.u64)
D0.u64 = saveexec.u64
SCC.b32 = EXEC.u64 != 0
# --- end pseudocode ---
def _SOP1Op_S_AND_NOT1_SAVEEXEC_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Calculate bitwise AND on the scalar input and the negation of the EXEC mask, store the calculated result into
# the EXEC mask, set SCC iff the calculated result is nonzero and store the original value of the EXEC mask into
# saveexec = EXEC.u32;
# EXEC.u32 = (S0.u32 & ~EXEC.u32);
# D0.u32 = saveexec.u32;
# SCC = EXEC.u32 != 0U
saveexec = Reg(EXEC._val)
# --- compiled pseudocode ---
saveexec = Reg(EXEC.u32)
EXEC.u32 = (S0.u32 & ~EXEC.u32)
D0.u32 = saveexec.u32
SCC.b32 = EXEC.u32 != 0
# --- end pseudocode ---
def _SOP1Op_S_AND_NOT1_SAVEEXEC_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Calculate bitwise AND on the scalar input and the negation of the EXEC mask, store the calculated result into
# the EXEC mask, set SCC iff the calculated result is nonzero and store the original value of the EXEC mask into
# saveexec = EXEC.u64;
# EXEC.u64 = (S0.u64 & ~EXEC.u64);
# D0.u64 = saveexec.u64;
# SCC = EXEC.u64 != 0ULL
saveexec = Reg(EXEC._val)
# --- compiled pseudocode ---
saveexec = Reg(EXEC.u64)
EXEC.u64 = (S0.u64 & ~EXEC.u64)
D0.u64 = saveexec.u64
SCC.b32 = EXEC.u64 != 0
# --- end pseudocode ---
def _SOP1Op_S_OR_NOT1_SAVEEXEC_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Calculate bitwise OR on the scalar input and the negation of the EXEC mask, store the calculated result into the
# EXEC mask, set SCC iff the calculated result is nonzero and store the original value of the EXEC mask into the
# saveexec = EXEC.u32;
# EXEC.u32 = (S0.u32 | ~EXEC.u32);
# D0.u32 = saveexec.u32;
# SCC = EXEC.u32 != 0U
saveexec = Reg(EXEC._val)
# --- compiled pseudocode ---
saveexec = Reg(EXEC.u32)
EXEC.u32 = (S0.u32 | ~EXEC.u32)
D0.u32 = saveexec.u32
SCC.b32 = EXEC.u32 != 0
# --- end pseudocode ---
def _SOP1Op_S_OR_NOT1_SAVEEXEC_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Calculate bitwise OR on the scalar input and the negation of the EXEC mask, store the calculated result into the
# EXEC mask, set SCC iff the calculated result is nonzero and store the original value of the EXEC mask into the
# saveexec = EXEC.u64;
# EXEC.u64 = (S0.u64 | ~EXEC.u64);
# D0.u64 = saveexec.u64;
# SCC = EXEC.u64 != 0ULL
saveexec = Reg(EXEC._val)
# --- compiled pseudocode ---
saveexec = Reg(EXEC.u64)
EXEC.u64 = (S0.u64 | ~EXEC.u64)
D0.u64 = saveexec.u64
SCC.b32 = EXEC.u64 != 0
# --- end pseudocode ---
def _SOP1Op_S_AND_NOT0_WREXEC_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Calculate bitwise AND on the EXEC mask and the negation of the scalar input, store the calculated result into
# Unlike the SAVEEXEC series of opcodes, the value written to destination SGPRs is the result of the bitwise-op
# result. EXEC and the destination SGPRs have the same value at the end of this instruction. This instruction is
# EXEC.u32 = (~S0.u32 & EXEC.u32);
# D0.u32 = EXEC.u32;
# SCC = EXEC.u32 != 0U
# --- compiled pseudocode ---
EXEC.u32 = (~S0.u32 & EXEC.u32)
D0.u32 = EXEC.u32
SCC.b32 = EXEC.u32 != 0
# --- end pseudocode ---
def _SOP1Op_S_AND_NOT0_WREXEC_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Calculate bitwise AND on the EXEC mask and the negation of the scalar input, store the calculated result into
# Unlike the SAVEEXEC series of opcodes, the value written to destination SGPRs is the result of the bitwise-op
# result. EXEC and the destination SGPRs have the same value at the end of this instruction. This instruction is
# EXEC.u64 = (~S0.u64 & EXEC.u64);
# D0.u64 = EXEC.u64;
# SCC = EXEC.u64 != 0ULL
# --- compiled pseudocode ---
EXEC.u64 = (~S0.u64 & EXEC.u64)
D0.u64 = EXEC.u64
SCC.b32 = EXEC.u64 != 0
# --- end pseudocode ---
def _SOP1Op_S_AND_NOT1_WREXEC_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Calculate bitwise AND on the scalar input and the negation of the EXEC mask, store the calculated result into
# Unlike the SAVEEXEC series of opcodes, the value written to destination SGPRs is the result of the bitwise-op
# result. EXEC and the destination SGPRs have the same value at the end of this instruction. This instruction is
# EXEC.u32 = (S0.u32 & ~EXEC.u32);
# D0.u32 = EXEC.u32;
# SCC = EXEC.u32 != 0U
# --- compiled pseudocode ---
EXEC.u32 = (S0.u32 & ~EXEC.u32)
D0.u32 = EXEC.u32
SCC.b32 = EXEC.u32 != 0
# --- end pseudocode ---
def _SOP1Op_S_AND_NOT1_WREXEC_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Calculate bitwise AND on the scalar input and the negation of the EXEC mask, store the calculated result into
# Unlike the SAVEEXEC series of opcodes, the value written to destination SGPRs is the result of the bitwise-op
# result. EXEC and the destination SGPRs have the same value at the end of this instruction. This instruction is
# EXEC.u64 = (S0.u64 & ~EXEC.u64);
# D0.u64 = EXEC.u64;
# SCC = EXEC.u64 != 0ULL
# --- compiled pseudocode ---
EXEC.u64 = (S0.u64 & ~EXEC.u64)
D0.u64 = EXEC.u64
SCC.b32 = EXEC.u64 != 0
# --- end pseudocode ---
def _SOP1Op_S_SENDMSG_RTN_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# If SDST is VCC then VCCZ is undefined.
# --- compiled pseudocode ---
# --- end pseudocode ---
pass
def _SOP1Op_S_SENDMSG_RTN_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# If SDST is VCC then VCCZ is undefined.
# --- compiled pseudocode ---
# --- end pseudocode ---
pass
def _SOP1Op_S_CEIL_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = trunc(S0.f32);
# if ((S0.f32 > 0.0F) && (S0.f32 != D0.f32)) then
# D0.f32 += 1.0F
# endif
# --- compiled pseudocode ---
D0.f32 = trunc(S0.f32)
if ((S0.f32 > 0.0) and (S0.f32 != D0.f32)):
D0.f32 += 1.0
# --- end pseudocode ---
def _SOP1Op_S_FLOOR_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = trunc(S0.f32);
# if ((S0.f32 < 0.0F) && (S0.f32 != D0.f32)) then
# D0.f32 += -1.0F
# endif
# --- compiled pseudocode ---
D0.f32 = trunc(S0.f32)
if ((S0.f32 < 0.0) and (S0.f32 != D0.f32)):
D0.f32 += -1.0
# --- end pseudocode ---
def _SOP1Op_S_TRUNC_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = trunc(S0.f32)
# --- compiled pseudocode ---
D0.f32 = trunc(S0.f32)
# --- end pseudocode ---
def _SOP1Op_S_RNDNE_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = floor(S0.f32 + 0.5F);
# if (isEven(64'F(floor(S0.f32))) && (fract(S0.f32) == 0.5F)) then
# D0.f32 -= 1.0F
# endif
# --- compiled pseudocode ---
D0.f32 = floor(S0.f32 + 0.5)
if (isEven(F(floor(S0.f32))) and (fract(S0.f32) == 0.5)):
D0.f32 -= 1.0
# --- end pseudocode ---
def _SOP1Op_S_CVT_F32_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = i32_to_f32(S0.i32)
# --- compiled pseudocode ---
D0.f32 = i32_to_f32(S0.i32)
# --- end pseudocode ---
def _SOP1Op_S_CVT_F32_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = u32_to_f32(S0.u32)
# --- compiled pseudocode ---
D0.f32 = u32_to_f32(S0.u32)
# --- end pseudocode ---
def _SOP1Op_S_CVT_I32_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = f32_to_i32(S0.f32)
# --- compiled pseudocode ---
D0.i32 = f32_to_i32(S0.f32)
# --- end pseudocode ---
def _SOP1Op_S_CVT_U32_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = f32_to_u32(S0.f32)
# --- compiled pseudocode ---
D0.u32 = f32_to_u32(S0.f32)
# --- end pseudocode ---
def _SOP1Op_S_CVT_F16_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = f32_to_f16(S0.f32)
# --- compiled pseudocode ---
D0.f16 = f32_to_f16(S0.f32)
# --- end pseudocode ---
def _SOP1Op_S_CVT_F32_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = f16_to_f32(S0.f16)
# --- compiled pseudocode ---
D0.f32 = f16_to_f32(S0.f16)
# --- end pseudocode ---
def _SOP1Op_S_CVT_HI_F32_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = f16_to_f32(S0[31 : 16].f16)
# --- compiled pseudocode ---
D0.f32 = f16_to_f32(S0[31 : 16].f16)
# --- end pseudocode ---
def _SOP1Op_S_CEIL_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = trunc(S0.f16);
# if ((S0.f16 > 16'0.0) && (S0.f16 != D0.f16)) then
# D0.f16 += 16'1.0
# endif
# --- compiled pseudocode ---
D0.f16 = trunc(S0.f16)
if ((S0.f16 > 0.0) and (S0.f16 != D0.f16)):
D0.f16 += 1.0
# --- end pseudocode ---
def _SOP1Op_S_FLOOR_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = trunc(S0.f16);
# if ((S0.f16 < 16'0.0) && (S0.f16 != D0.f16)) then
# D0.f16 += -16'1.0
# endif
# --- compiled pseudocode ---
D0.f16 = trunc(S0.f16)
if ((S0.f16 < 0.0) and (S0.f16 != D0.f16)):
D0.f16 += -1.0
# --- end pseudocode ---
def _SOP1Op_S_TRUNC_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = trunc(S0.f16)
# --- compiled pseudocode ---
D0.f16 = trunc(S0.f16)
# --- end pseudocode ---
def _SOP1Op_S_RNDNE_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = floor(S0.f16 + 16'0.5);
# if (isEven(64'F(floor(S0.f16))) && (fract(S0.f16) == 16'0.5)) then
# D0.f16 -= 16'1.0
# endif
# --- compiled pseudocode ---
D0.f16 = floor(S0.f16 + 0.5)
if (isEven(F(floor(S0.f16))) and (fract(S0.f16) == 0.5)):
D0.f16 -= 1.0
# --- end pseudocode ---
SOP1Op_FUNCTIONS = {
SOP1Op.S_MOV_B32: _SOP1Op_S_MOV_B32,
SOP1Op.S_MOV_B64: _SOP1Op_S_MOV_B64,
SOP1Op.S_CMOV_B32: _SOP1Op_S_CMOV_B32,
SOP1Op.S_CMOV_B64: _SOP1Op_S_CMOV_B64,
SOP1Op.S_BREV_B32: _SOP1Op_S_BREV_B32,
SOP1Op.S_BREV_B64: _SOP1Op_S_BREV_B64,
SOP1Op.S_CTZ_I32_B32: _SOP1Op_S_CTZ_I32_B32,
SOP1Op.S_CTZ_I32_B64: _SOP1Op_S_CTZ_I32_B64,
SOP1Op.S_CLZ_I32_U32: _SOP1Op_S_CLZ_I32_U32,
SOP1Op.S_CLZ_I32_U64: _SOP1Op_S_CLZ_I32_U64,
SOP1Op.S_CLS_I32: _SOP1Op_S_CLS_I32,
SOP1Op.S_CLS_I32_I64: _SOP1Op_S_CLS_I32_I64,
SOP1Op.S_SEXT_I32_I8: _SOP1Op_S_SEXT_I32_I8,
SOP1Op.S_SEXT_I32_I16: _SOP1Op_S_SEXT_I32_I16,
SOP1Op.S_BITSET0_B32: _SOP1Op_S_BITSET0_B32,
SOP1Op.S_BITSET0_B64: _SOP1Op_S_BITSET0_B64,
SOP1Op.S_BITSET1_B32: _SOP1Op_S_BITSET1_B32,
SOP1Op.S_BITSET1_B64: _SOP1Op_S_BITSET1_B64,
SOP1Op.S_BITREPLICATE_B64_B32: _SOP1Op_S_BITREPLICATE_B64_B32,
SOP1Op.S_ABS_I32: _SOP1Op_S_ABS_I32,
SOP1Op.S_BCNT0_I32_B32: _SOP1Op_S_BCNT0_I32_B32,
SOP1Op.S_BCNT0_I32_B64: _SOP1Op_S_BCNT0_I32_B64,
SOP1Op.S_BCNT1_I32_B32: _SOP1Op_S_BCNT1_I32_B32,
SOP1Op.S_BCNT1_I32_B64: _SOP1Op_S_BCNT1_I32_B64,
SOP1Op.S_QUADMASK_B32: _SOP1Op_S_QUADMASK_B32,
SOP1Op.S_QUADMASK_B64: _SOP1Op_S_QUADMASK_B64,
SOP1Op.S_WQM_B32: _SOP1Op_S_WQM_B32,
SOP1Op.S_WQM_B64: _SOP1Op_S_WQM_B64,
SOP1Op.S_NOT_B32: _SOP1Op_S_NOT_B32,
SOP1Op.S_NOT_B64: _SOP1Op_S_NOT_B64,
SOP1Op.S_AND_SAVEEXEC_B32: _SOP1Op_S_AND_SAVEEXEC_B32,
SOP1Op.S_AND_SAVEEXEC_B64: _SOP1Op_S_AND_SAVEEXEC_B64,
SOP1Op.S_OR_SAVEEXEC_B32: _SOP1Op_S_OR_SAVEEXEC_B32,
SOP1Op.S_OR_SAVEEXEC_B64: _SOP1Op_S_OR_SAVEEXEC_B64,
SOP1Op.S_XOR_SAVEEXEC_B32: _SOP1Op_S_XOR_SAVEEXEC_B32,
SOP1Op.S_XOR_SAVEEXEC_B64: _SOP1Op_S_XOR_SAVEEXEC_B64,
SOP1Op.S_NAND_SAVEEXEC_B32: _SOP1Op_S_NAND_SAVEEXEC_B32,
SOP1Op.S_NAND_SAVEEXEC_B64: _SOP1Op_S_NAND_SAVEEXEC_B64,
SOP1Op.S_NOR_SAVEEXEC_B32: _SOP1Op_S_NOR_SAVEEXEC_B32,
SOP1Op.S_NOR_SAVEEXEC_B64: _SOP1Op_S_NOR_SAVEEXEC_B64,
SOP1Op.S_XNOR_SAVEEXEC_B32: _SOP1Op_S_XNOR_SAVEEXEC_B32,
SOP1Op.S_XNOR_SAVEEXEC_B64: _SOP1Op_S_XNOR_SAVEEXEC_B64,
SOP1Op.S_AND_NOT0_SAVEEXEC_B32: _SOP1Op_S_AND_NOT0_SAVEEXEC_B32,
SOP1Op.S_AND_NOT0_SAVEEXEC_B64: _SOP1Op_S_AND_NOT0_SAVEEXEC_B64,
SOP1Op.S_OR_NOT0_SAVEEXEC_B32: _SOP1Op_S_OR_NOT0_SAVEEXEC_B32,
SOP1Op.S_OR_NOT0_SAVEEXEC_B64: _SOP1Op_S_OR_NOT0_SAVEEXEC_B64,
SOP1Op.S_AND_NOT1_SAVEEXEC_B32: _SOP1Op_S_AND_NOT1_SAVEEXEC_B32,
SOP1Op.S_AND_NOT1_SAVEEXEC_B64: _SOP1Op_S_AND_NOT1_SAVEEXEC_B64,
SOP1Op.S_OR_NOT1_SAVEEXEC_B32: _SOP1Op_S_OR_NOT1_SAVEEXEC_B32,
SOP1Op.S_OR_NOT1_SAVEEXEC_B64: _SOP1Op_S_OR_NOT1_SAVEEXEC_B64,
SOP1Op.S_AND_NOT0_WREXEC_B32: _SOP1Op_S_AND_NOT0_WREXEC_B32,
SOP1Op.S_AND_NOT0_WREXEC_B64: _SOP1Op_S_AND_NOT0_WREXEC_B64,
SOP1Op.S_AND_NOT1_WREXEC_B32: _SOP1Op_S_AND_NOT1_WREXEC_B32,
SOP1Op.S_AND_NOT1_WREXEC_B64: _SOP1Op_S_AND_NOT1_WREXEC_B64,
SOP1Op.S_SENDMSG_RTN_B32: _SOP1Op_S_SENDMSG_RTN_B32,
SOP1Op.S_SENDMSG_RTN_B64: _SOP1Op_S_SENDMSG_RTN_B64,
SOP1Op.S_CEIL_F32: _SOP1Op_S_CEIL_F32,
SOP1Op.S_FLOOR_F32: _SOP1Op_S_FLOOR_F32,
SOP1Op.S_TRUNC_F32: _SOP1Op_S_TRUNC_F32,
SOP1Op.S_RNDNE_F32: _SOP1Op_S_RNDNE_F32,
SOP1Op.S_CVT_F32_I32: _SOP1Op_S_CVT_F32_I32,
SOP1Op.S_CVT_F32_U32: _SOP1Op_S_CVT_F32_U32,
SOP1Op.S_CVT_I32_F32: _SOP1Op_S_CVT_I32_F32,
SOP1Op.S_CVT_U32_F32: _SOP1Op_S_CVT_U32_F32,
SOP1Op.S_CVT_F16_F32: _SOP1Op_S_CVT_F16_F32,
SOP1Op.S_CVT_F32_F16: _SOP1Op_S_CVT_F32_F16,
SOP1Op.S_CVT_HI_F32_F16: _SOP1Op_S_CVT_HI_F32_F16,
SOP1Op.S_CEIL_F16: _SOP1Op_S_CEIL_F16,
SOP1Op.S_FLOOR_F16: _SOP1Op_S_FLOOR_F16,
SOP1Op.S_TRUNC_F16: _SOP1Op_S_TRUNC_F16,
SOP1Op.S_RNDNE_F16: _SOP1Op_S_RNDNE_F16,
}
def _SOP2Op_S_ADD_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = 64'U(S0.u32) + 64'U(S1.u32);
# SCC = tmp >= 0x100000000ULL ? 1'1U : 1'0U;
# D0.u32 = tmp.u32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg((S0.u32) + (S1.u32))
SCC.b32 = ((1) if (tmp >= 0x100000000) else (0))
D0.u32 = tmp.u32
# --- end pseudocode ---
def _SOP2Op_S_SUB_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = S0.u32 - S1.u32;
# SCC = S1.u32 > S0.u32 ? 1'1U : 1'0U;
# D0.u32 = tmp.u32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(S0.u32 - S1.u32)
SCC.b32 = ((1) if (S1.u32 > S0.u32) else (0))
D0.u32 = tmp.u32
# --- end pseudocode ---
def _SOP2Op_S_ADD_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = S0.i32 + S1.i32;
# SCC = ((S0.u32[31] == S1.u32[31]) && (S0.u32[31] != tmp.u32[31]));
# D0.i32 = tmp.i32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(S0.i32 + S1.i32)
SCC.b32 = ((S0.u32[31] == S1.u32[31]) and (S0.u32[31] != tmp.u32[31]))
D0.i32 = tmp.i32
# --- end pseudocode ---
def _SOP2Op_S_SUB_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = S0.i32 - S1.i32;
# SCC = ((S0.u32[31] != S1.u32[31]) && (S0.u32[31] != tmp.u32[31]));
# D0.i32 = tmp.i32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(S0.i32 - S1.i32)
SCC.b32 = ((S0.u32[31] != S1.u32[31]) and (S0.u32[31] != tmp.u32[31]))
D0.i32 = tmp.i32
# --- end pseudocode ---
def _SOP2Op_S_ADDC_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = 64'U(S0.u32) + 64'U(S1.u32) + SCC.u64;
# SCC = tmp >= 0x100000000ULL ? 1'1U : 1'0U;
# D0.u32 = tmp.u32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg((S0.u32) + (S1.u32) + SCC.u64)
SCC.b32 = ((1) if (tmp >= 0x100000000) else (0))
D0.u32 = tmp.u32
# --- end pseudocode ---
def _SOP2Op_S_SUBB_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = S0.u32 - S1.u32 - SCC.u32;
# SCC = 64'U(S1.u32) + SCC.u64 > 64'U(S0.u32) ? 1'1U : 1'0U;
# D0.u32 = tmp.u32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(S0.u32 - S1.u32 - SCC.u32)
SCC.b32 = ((1) if ((S1.u32) + SCC.u64 > (S0.u32)) else (0))
D0.u32 = tmp.u32
# --- end pseudocode ---
def _SOP2Op_S_ABSDIFF_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = S0.i32 - S1.i32;
# if D0.i32 < 0 then
# D0.i32 = -D0.i32
# endif;
# SCC = D0.i32 != 0
# --- compiled pseudocode ---
D0.i32 = S0.i32 - S1.i32
if D0.i32 < 0:
D0.i32 = -D0.i32
SCC.b32 = D0.i32 != 0
# --- end pseudocode ---
def _SOP2Op_S_LSHL_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = (S0.u32 << S1[4 : 0].u32);
# SCC = D0.u32 != 0U
# --- compiled pseudocode ---
D0.u32 = (S0.u32 << S1[4 : 0].u32)
SCC.b32 = D0.u32 != 0
# --- end pseudocode ---
def _SOP2Op_S_LSHL_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64 = (S0.u64 << S1[5 : 0].u32);
# SCC = D0.u64 != 0ULL
# --- compiled pseudocode ---
D0.u64 = (S0.u64 << S1[5 : 0].u32)
SCC.b32 = D0.u64 != 0
# --- end pseudocode ---
def _SOP2Op_S_LSHR_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = (S0.u32 >> S1[4 : 0].u32);
# SCC = D0.u32 != 0U
# --- compiled pseudocode ---
D0.u32 = (S0.u32 >> S1[4 : 0].u32)
SCC.b32 = D0.u32 != 0
# --- end pseudocode ---
def _SOP2Op_S_LSHR_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64 = (S0.u64 >> S1[5 : 0].u32);
# SCC = D0.u64 != 0ULL
# --- compiled pseudocode ---
D0.u64 = (S0.u64 >> S1[5 : 0].u32)
SCC.b32 = D0.u64 != 0
# --- end pseudocode ---
def _SOP2Op_S_ASHR_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = 32'I(signext(S0.i32) >> S1[4 : 0].u32);
# SCC = D0.i32 != 0
# --- compiled pseudocode ---
D0.i32 = (signext(S0.i32) >> S1[4 : 0].u32)
SCC.b32 = D0.i32 != 0
# --- end pseudocode ---
def _SOP2Op_S_ASHR_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i64 = (signext(S0.i64) >> S1[5 : 0].u32);
# SCC = D0.i64 != 0LL
# --- compiled pseudocode ---
D0.i64 = (signext(S0.i64) >> S1[5 : 0].u32)
SCC.b32 = D0.i64 != 0
# --- end pseudocode ---
def _SOP2Op_S_LSHL1_ADD_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = (64'U(S0.u32) << 1U) + 64'U(S1.u32);
# SCC = tmp >= 0x100000000ULL ? 1'1U : 1'0U;
# D0.u32 = tmp.u32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(((S0.u32) << 1) + (S1.u32))
SCC.b32 = ((1) if (tmp >= 0x100000000) else (0))
D0.u32 = tmp.u32
# --- end pseudocode ---
def _SOP2Op_S_LSHL2_ADD_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = (64'U(S0.u32) << 2U) + 64'U(S1.u32);
# SCC = tmp >= 0x100000000ULL ? 1'1U : 1'0U;
# D0.u32 = tmp.u32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(((S0.u32) << 2) + (S1.u32))
SCC.b32 = ((1) if (tmp >= 0x100000000) else (0))
D0.u32 = tmp.u32
# --- end pseudocode ---
def _SOP2Op_S_LSHL3_ADD_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = (64'U(S0.u32) << 3U) + 64'U(S1.u32);
# SCC = tmp >= 0x100000000ULL ? 1'1U : 1'0U;
# D0.u32 = tmp.u32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(((S0.u32) << 3) + (S1.u32))
SCC.b32 = ((1) if (tmp >= 0x100000000) else (0))
D0.u32 = tmp.u32
# --- end pseudocode ---
def _SOP2Op_S_LSHL4_ADD_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = (64'U(S0.u32) << 4U) + 64'U(S1.u32);
# SCC = tmp >= 0x100000000ULL ? 1'1U : 1'0U;
# D0.u32 = tmp.u32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(((S0.u32) << 4) + (S1.u32))
SCC.b32 = ((1) if (tmp >= 0x100000000) else (0))
D0.u32 = tmp.u32
# --- end pseudocode ---
def _SOP2Op_S_MIN_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.i32 < S1.i32;
# D0.i32 = SCC ? S0.i32 : S1.i32
# --- compiled pseudocode ---
SCC.b32 = S0.i32 < S1.i32
D0.i32 = ((S0.i32) if (SCC) else (S1.i32))
# --- end pseudocode ---
def _SOP2Op_S_MIN_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.u32 < S1.u32;
# D0.u32 = SCC ? S0.u32 : S1.u32
# --- compiled pseudocode ---
SCC.b32 = S0.u32 < S1.u32
D0.u32 = ((S0.u32) if (SCC) else (S1.u32))
# --- end pseudocode ---
def _SOP2Op_S_MAX_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.i32 >= S1.i32;
# D0.i32 = SCC ? S0.i32 : S1.i32
# --- compiled pseudocode ---
SCC.b32 = S0.i32 >= S1.i32
D0.i32 = ((S0.i32) if (SCC) else (S1.i32))
# --- end pseudocode ---
def _SOP2Op_S_MAX_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.u32 >= S1.u32;
# D0.u32 = SCC ? S0.u32 : S1.u32
# --- compiled pseudocode ---
SCC.b32 = S0.u32 >= S1.u32
D0.u32 = ((S0.u32) if (SCC) else (S1.u32))
# --- end pseudocode ---
def _SOP2Op_S_AND_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = (S0.u32 & S1.u32);
# SCC = D0.u32 != 0U
# --- compiled pseudocode ---
D0.u32 = (S0.u32 & S1.u32)
SCC.b32 = D0.u32 != 0
# --- end pseudocode ---
def _SOP2Op_S_AND_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64 = (S0.u64 & S1.u64);
# SCC = D0.u64 != 0ULL
# --- compiled pseudocode ---
D0.u64 = (S0.u64 & S1.u64)
SCC.b32 = D0.u64 != 0
# --- end pseudocode ---
def _SOP2Op_S_OR_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = (S0.u32 | S1.u32);
# SCC = D0.u32 != 0U
# --- compiled pseudocode ---
D0.u32 = (S0.u32 | S1.u32)
SCC.b32 = D0.u32 != 0
# --- end pseudocode ---
def _SOP2Op_S_OR_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64 = (S0.u64 | S1.u64);
# SCC = D0.u64 != 0ULL
# --- compiled pseudocode ---
D0.u64 = (S0.u64 | S1.u64)
SCC.b32 = D0.u64 != 0
# --- end pseudocode ---
def _SOP2Op_S_XOR_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = (S0.u32 ^ S1.u32);
# SCC = D0.u32 != 0U
# --- compiled pseudocode ---
D0.u32 = (S0.u32 ^ S1.u32)
SCC.b32 = D0.u32 != 0
# --- end pseudocode ---
def _SOP2Op_S_XOR_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64 = (S0.u64 ^ S1.u64);
# SCC = D0.u64 != 0ULL
# --- compiled pseudocode ---
D0.u64 = (S0.u64 ^ S1.u64)
SCC.b32 = D0.u64 != 0
# --- end pseudocode ---
def _SOP2Op_S_NAND_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = ~(S0.u32 & S1.u32);
# SCC = D0.u32 != 0U
# --- compiled pseudocode ---
D0.u32 = ~(S0.u32 & S1.u32)
SCC.b32 = D0.u32 != 0
# --- end pseudocode ---
def _SOP2Op_S_NAND_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64 = ~(S0.u64 & S1.u64);
# SCC = D0.u64 != 0ULL
# --- compiled pseudocode ---
D0.u64 = ~(S0.u64 & S1.u64)
SCC.b32 = D0.u64 != 0
# --- end pseudocode ---
def _SOP2Op_S_NOR_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = ~(S0.u32 | S1.u32);
# SCC = D0.u32 != 0U
# --- compiled pseudocode ---
D0.u32 = ~(S0.u32 | S1.u32)
SCC.b32 = D0.u32 != 0
# --- end pseudocode ---
def _SOP2Op_S_NOR_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64 = ~(S0.u64 | S1.u64);
# SCC = D0.u64 != 0ULL
# --- compiled pseudocode ---
D0.u64 = ~(S0.u64 | S1.u64)
SCC.b32 = D0.u64 != 0
# --- end pseudocode ---
def _SOP2Op_S_XNOR_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = ~(S0.u32 ^ S1.u32);
# SCC = D0.u32 != 0U
# --- compiled pseudocode ---
D0.u32 = ~(S0.u32 ^ S1.u32)
SCC.b32 = D0.u32 != 0
# --- end pseudocode ---
def _SOP2Op_S_XNOR_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64 = ~(S0.u64 ^ S1.u64);
# SCC = D0.u64 != 0ULL
# --- compiled pseudocode ---
D0.u64 = ~(S0.u64 ^ S1.u64)
SCC.b32 = D0.u64 != 0
# --- end pseudocode ---
def _SOP2Op_S_AND_NOT1_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = (S0.u32 & ~S1.u32);
# SCC = D0.u32 != 0U
# --- compiled pseudocode ---
D0.u32 = (S0.u32 & ~S1.u32)
SCC.b32 = D0.u32 != 0
# --- end pseudocode ---
def _SOP2Op_S_AND_NOT1_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64 = (S0.u64 & ~S1.u64);
# SCC = D0.u64 != 0ULL
# --- compiled pseudocode ---
D0.u64 = (S0.u64 & ~S1.u64)
SCC.b32 = D0.u64 != 0
# --- end pseudocode ---
def _SOP2Op_S_OR_NOT1_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = (S0.u32 | ~S1.u32);
# SCC = D0.u32 != 0U
# --- compiled pseudocode ---
D0.u32 = (S0.u32 | ~S1.u32)
SCC.b32 = D0.u32 != 0
# --- end pseudocode ---
def _SOP2Op_S_OR_NOT1_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64 = (S0.u64 | ~S1.u64);
# SCC = D0.u64 != 0ULL
# --- compiled pseudocode ---
D0.u64 = (S0.u64 | ~S1.u64)
SCC.b32 = D0.u64 != 0
# --- end pseudocode ---
def _SOP2Op_S_BFE_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = ((S0.u32 >> S1[4 : 0].u32) & ((1U << S1[22 : 16].u32) - 1U));
# SCC = D0.u32 != 0U
# --- compiled pseudocode ---
D0.u32 = ((S0.u32 >> S1[4 : 0].u32) & ((1 << S1[22 : 16].u32) - 1))
SCC.b32 = D0.u32 != 0
# --- end pseudocode ---
def _SOP2Op_S_BFE_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp.i32 = ((S0.i32 >> S1[4 : 0].u32) & ((1 << S1[22 : 16].u32) - 1));
# D0.i32 = signext_from_bit(tmp.i32, S1[22 : 16].u32);
# SCC = D0.i32 != 0
tmp = Reg(0)
# --- compiled pseudocode ---
tmp.i32 = ((S0.i32 >> S1[4 : 0].u32) & ((1 << S1[22 : 16].u32) - 1))
D0.i32 = signext_from_bit(tmp.i32, S1[22 : 16].u32)
SCC.b32 = D0.i32 != 0
# --- end pseudocode ---
def _SOP2Op_S_BFE_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64 = ((S0.u64 >> S1[5 : 0].u32) & ((1ULL << S1[22 : 16].u32) - 1ULL));
# SCC = D0.u64 != 0ULL
# --- compiled pseudocode ---
D0.u64 = ((S0.u64 >> S1[5 : 0].u32) & ((1 << S1[22 : 16].u32) - 1))
SCC.b32 = D0.u64 != 0
# --- end pseudocode ---
def _SOP2Op_S_BFE_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp.i64 = ((S0.i64 >> S1[5 : 0].u32) & ((1LL << S1[22 : 16].u32) - 1LL));
# D0.i64 = signext_from_bit(tmp.i64, S1[22 : 16].u32);
# SCC = D0.i64 != 0LL
tmp = Reg(0)
# --- compiled pseudocode ---
tmp.i64 = ((S0.i64 >> S1[5 : 0].u32) & ((1 << S1[22 : 16].u32) - 1))
D0.i64 = signext_from_bit(tmp.i64, S1[22 : 16].u32)
SCC.b32 = D0.i64 != 0
# --- end pseudocode ---
def _SOP2Op_S_BFM_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = (((1U << S0[4 : 0].u32) - 1U) << S1[4 : 0].u32)
# --- compiled pseudocode ---
D0.u32 = (((1 << S0[4 : 0].u32) - 1) << S1[4 : 0].u32)
# --- end pseudocode ---
def _SOP2Op_S_BFM_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64 = (((1ULL << S0[5 : 0].u32) - 1ULL) << S1[5 : 0].u32)
# --- compiled pseudocode ---
D0.u64 = (((1 << S0[5 : 0].u32) - 1) << S1[5 : 0].u32)
# --- end pseudocode ---
def _SOP2Op_S_MUL_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = S0.i32 * S1.i32
# --- compiled pseudocode ---
D0.i32 = S0.i32 * S1.i32
# --- end pseudocode ---
def _SOP2Op_S_MUL_HI_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = 32'U((64'U(S0.u32) * 64'U(S1.u32)) >> 32U)
# --- compiled pseudocode ---
D0.u32 = (((S0.u32) * (S1.u32)) >> 32)
# --- end pseudocode ---
def _SOP2Op_S_MUL_HI_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = 32'I((64'I(S0.i32) * 64'I(S1.i32)) >> 32U)
# --- compiled pseudocode ---
D0.i32 = (((S0.i32) * (S1.i32)) >> 32)
# --- end pseudocode ---
def _SOP2Op_S_CSELECT_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = SCC ? S0.u32 : S1.u32
# --- compiled pseudocode ---
D0.u32 = ((S0.u32) if (SCC) else (S1.u32))
# --- end pseudocode ---
def _SOP2Op_S_CSELECT_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64 = SCC ? S0.u64 : S1.u64
# --- compiled pseudocode ---
D0.u64 = ((S0.u64) if (SCC) else (S1.u64))
# --- end pseudocode ---
def _SOP2Op_S_PACK_LL_B32_B16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0 = { S1[15 : 0].u16, S0[15 : 0].u16 }
# --- compiled pseudocode ---
D0.b32 = _pack(S1[15 : 0].u16, S0[15 : 0].u16)
# --- end pseudocode ---
def _SOP2Op_S_PACK_LH_B32_B16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0 = { S1[31 : 16].u16, S0[15 : 0].u16 }
# --- compiled pseudocode ---
D0.b32 = _pack(S1[31 : 16].u16, S0[15 : 0].u16)
# --- end pseudocode ---
def _SOP2Op_S_PACK_HH_B32_B16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0 = { S1[31 : 16].u16, S0[31 : 16].u16 }
# --- compiled pseudocode ---
D0.b32 = _pack(S1[31 : 16].u16, S0[31 : 16].u16)
# --- end pseudocode ---
def _SOP2Op_S_PACK_HL_B32_B16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0 = { S1[15 : 0].u16, S0[31 : 16].u16 }
# --- compiled pseudocode ---
D0.b32 = _pack(S1[15 : 0].u16, S0[31 : 16].u16)
# --- end pseudocode ---
def _SOP2Op_S_ADD_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = S0.f32 + S1.f32
# --- compiled pseudocode ---
D0.f32 = S0.f32 + S1.f32
# --- end pseudocode ---
def _SOP2Op_S_SUB_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = S0.f32 - S1.f32
# --- compiled pseudocode ---
D0.f32 = S0.f32 - S1.f32
# --- end pseudocode ---
def _SOP2Op_S_MIN_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# // Version of comparison where -0.0 < +0.0, differs from IEEE
# if WAVE_MODE.IEEE then
# if isSignalNAN(64'F(S0.f32)) then
# D0.f32 = 32'F(cvtToQuietNAN(64'F(S0.f32)))
# elsif isSignalNAN(64'F(S1.f32)) then
# D0.f32 = 32'F(cvtToQuietNAN(64'F(S1.f32)))
# elsif isQuietNAN(64'F(S1.f32)) then
# D0.f32 = S0.f32
# elsif isQuietNAN(64'F(S0.f32)) then
# D0.f32 = S1.f32
# elsif LT_NEG_ZERO(S0.f32, S1.f32) then
# // NOTE: -0<+0 is TRUE in this comparison
# D0.f32 = S0.f32
# else
# D0.f32 = S1.f32
# endif
# else
# if isNAN(64'F(S1.f32)) then
# D0.f32 = S0.f32
# elsif isNAN(64'F(S0.f32)) then
# D0.f32 = S1.f32
# elsif LT_NEG_ZERO(S0.f32, S1.f32) then
# // NOTE: -0<+0 is TRUE in this comparison
# D0.f32 = S0.f32
# else
# D0.f32 = S1.f32
# endif
# endif;
# // Inequalities in the above pseudocode behave differently from IEEE
# --- compiled pseudocode ---
if WAVE_MODE.IEEE:
if isSignalNAN(F(S0.f32)):
D0.f32 = F(cvtToQuietNAN(F(S0.f32)))
elif isSignalNAN(F(S1.f32)):
D0.f32 = F(cvtToQuietNAN(F(S1.f32)))
elif isQuietNAN(F(S1.f32)):
D0.f32 = S0.f32
elif isQuietNAN(F(S0.f32)):
D0.f32 = S1.f32
elif LT_NEG_ZERO(S0.f32, S1.f32):
D0.f32 = S0.f32
else:
D0.f32 = S1.f32
else:
if isNAN(F(S1.f32)):
D0.f32 = S0.f32
elif isNAN(F(S0.f32)):
D0.f32 = S1.f32
elif LT_NEG_ZERO(S0.f32, S1.f32):
D0.f32 = S0.f32
else:
D0.f32 = S1.f32
# --- end pseudocode ---
def _SOP2Op_S_MAX_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# // Version of comparison where +0.0 > -0.0, differs from IEEE
# if WAVE_MODE.IEEE then
# if isSignalNAN(64'F(S0.f32)) then
# D0.f32 = 32'F(cvtToQuietNAN(64'F(S0.f32)))
# elsif isSignalNAN(64'F(S1.f32)) then
# D0.f32 = 32'F(cvtToQuietNAN(64'F(S1.f32)))
# elsif isQuietNAN(64'F(S1.f32)) then
# D0.f32 = S0.f32
# elsif isQuietNAN(64'F(S0.f32)) then
# D0.f32 = S1.f32
# elsif GT_NEG_ZERO(S0.f32, S1.f32) then
# // NOTE: +0>-0 is TRUE in this comparison
# D0.f32 = S0.f32
# else
# D0.f32 = S1.f32
# endif
# else
# if isNAN(64'F(S1.f32)) then
# D0.f32 = S0.f32
# elsif isNAN(64'F(S0.f32)) then
# D0.f32 = S1.f32
# elsif GT_NEG_ZERO(S0.f32, S1.f32) then
# // NOTE: +0>-0 is TRUE in this comparison
# D0.f32 = S0.f32
# else
# D0.f32 = S1.f32
# endif
# endif;
# // Inequalities in the above pseudocode behave differently from IEEE
# --- compiled pseudocode ---
if WAVE_MODE.IEEE:
if isSignalNAN(F(S0.f32)):
D0.f32 = F(cvtToQuietNAN(F(S0.f32)))
elif isSignalNAN(F(S1.f32)):
D0.f32 = F(cvtToQuietNAN(F(S1.f32)))
elif isQuietNAN(F(S1.f32)):
D0.f32 = S0.f32
elif isQuietNAN(F(S0.f32)):
D0.f32 = S1.f32
elif GT_NEG_ZERO(S0.f32, S1.f32):
D0.f32 = S0.f32
else:
D0.f32 = S1.f32
else:
if isNAN(F(S1.f32)):
D0.f32 = S0.f32
elif isNAN(F(S0.f32)):
D0.f32 = S1.f32
elif GT_NEG_ZERO(S0.f32, S1.f32):
D0.f32 = S0.f32
else:
D0.f32 = S1.f32
# --- end pseudocode ---
def _SOP2Op_S_MUL_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = S0.f32 * S1.f32
# --- compiled pseudocode ---
D0.f32 = S0.f32 * S1.f32
# --- end pseudocode ---
def _SOP2Op_S_FMAAK_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = fma(S0.f32, S1.f32, SIMM32.f32)
SIMM32 = SIMM16
# --- compiled pseudocode ---
D0.f32 = fma(S0.f32, S1.f32, SIMM32.f32)
# --- end pseudocode ---
def _SOP2Op_S_FMAMK_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = fma(S0.f32, SIMM32.f32, S1.f32)
SIMM32 = SIMM16
# --- compiled pseudocode ---
D0.f32 = fma(S0.f32, SIMM32.f32, S1.f32)
# --- end pseudocode ---
def _SOP2Op_S_FMAC_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = fma(S0.f32, S1.f32, D0.f32)
# --- compiled pseudocode ---
D0.f32 = fma(S0.f32, S1.f32, D0.f32)
# --- end pseudocode ---
def _SOP2Op_S_CVT_PK_RTZ_F16_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# prev_mode = ROUND_MODE;
# tmp[15 : 0].f16 = f32_to_f16(S0.f32);
# tmp[31 : 16].f16 = f32_to_f16(S1.f32);
tmp = Reg(0)
# --- compiled pseudocode ---
prev_mode = ROUND_MODE
tmp[15 : 0].f16 = f32_to_f16(S0.f32)
tmp[31 : 16].f16 = f32_to_f16(S1.f32)
# --- end pseudocode ---
def _SOP2Op_S_ADD_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = S0.f16 + S1.f16
# --- compiled pseudocode ---
D0.f16 = S0.f16 + S1.f16
# --- end pseudocode ---
def _SOP2Op_S_SUB_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = S0.f16 - S1.f16
# --- compiled pseudocode ---
D0.f16 = S0.f16 - S1.f16
# --- end pseudocode ---
def _SOP2Op_S_MIN_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# // Version of comparison where -0.0 < +0.0, differs from IEEE
# if WAVE_MODE.IEEE then
# if isSignalNAN(64'F(S0.f16)) then
# D0.f16 = 16'F(cvtToQuietNAN(64'F(S0.f16)))
# elsif isSignalNAN(64'F(S1.f16)) then
# D0.f16 = 16'F(cvtToQuietNAN(64'F(S1.f16)))
# elsif isQuietNAN(64'F(S1.f16)) then
# D0.f16 = S0.f16
# elsif isQuietNAN(64'F(S0.f16)) then
# D0.f16 = S1.f16
# elsif LT_NEG_ZERO(S0.f16, S1.f16) then
# // NOTE: -0<+0 is TRUE in this comparison
# D0.f16 = S0.f16
# else
# D0.f16 = S1.f16
# endif
# else
# if isNAN(64'F(S1.f16)) then
# D0.f16 = S0.f16
# elsif isNAN(64'F(S0.f16)) then
# D0.f16 = S1.f16
# elsif LT_NEG_ZERO(S0.f16, S1.f16) then
# // NOTE: -0<+0 is TRUE in this comparison
# D0.f16 = S0.f16
# else
# D0.f16 = S1.f16
# endif
# endif;
# // Inequalities in the above pseudocode behave differently from IEEE
# --- compiled pseudocode ---
if WAVE_MODE.IEEE:
if isSignalNAN(F(S0.f16)):
D0.f16 = F(cvtToQuietNAN(F(S0.f16)))
elif isSignalNAN(F(S1.f16)):
D0.f16 = F(cvtToQuietNAN(F(S1.f16)))
elif isQuietNAN(F(S1.f16)):
D0.f16 = S0.f16
elif isQuietNAN(F(S0.f16)):
D0.f16 = S1.f16
elif LT_NEG_ZERO(S0.f16, S1.f16):
D0.f16 = S0.f16
else:
D0.f16 = S1.f16
else:
if isNAN(F(S1.f16)):
D0.f16 = S0.f16
elif isNAN(F(S0.f16)):
D0.f16 = S1.f16
elif LT_NEG_ZERO(S0.f16, S1.f16):
D0.f16 = S0.f16
else:
D0.f16 = S1.f16
# --- end pseudocode ---
def _SOP2Op_S_MAX_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# // Version of comparison where +0.0 > -0.0, differs from IEEE
# if WAVE_MODE.IEEE then
# if isSignalNAN(64'F(S0.f16)) then
# D0.f16 = 16'F(cvtToQuietNAN(64'F(S0.f16)))
# elsif isSignalNAN(64'F(S1.f16)) then
# D0.f16 = 16'F(cvtToQuietNAN(64'F(S1.f16)))
# elsif isQuietNAN(64'F(S1.f16)) then
# D0.f16 = S0.f16
# elsif isQuietNAN(64'F(S0.f16)) then
# D0.f16 = S1.f16
# elsif GT_NEG_ZERO(S0.f16, S1.f16) then
# // NOTE: +0>-0 is TRUE in this comparison
# D0.f16 = S0.f16
# else
# D0.f16 = S1.f16
# endif
# else
# if isNAN(64'F(S1.f16)) then
# D0.f16 = S0.f16
# elsif isNAN(64'F(S0.f16)) then
# D0.f16 = S1.f16
# elsif GT_NEG_ZERO(S0.f16, S1.f16) then
# // NOTE: +0>-0 is TRUE in this comparison
# D0.f16 = S0.f16
# else
# D0.f16 = S1.f16
# endif
# endif;
# // Inequalities in the above pseudocode behave differently from IEEE
# --- compiled pseudocode ---
if WAVE_MODE.IEEE:
if isSignalNAN(F(S0.f16)):
D0.f16 = F(cvtToQuietNAN(F(S0.f16)))
elif isSignalNAN(F(S1.f16)):
D0.f16 = F(cvtToQuietNAN(F(S1.f16)))
elif isQuietNAN(F(S1.f16)):
D0.f16 = S0.f16
elif isQuietNAN(F(S0.f16)):
D0.f16 = S1.f16
elif GT_NEG_ZERO(S0.f16, S1.f16):
D0.f16 = S0.f16
else:
D0.f16 = S1.f16
else:
if isNAN(F(S1.f16)):
D0.f16 = S0.f16
elif isNAN(F(S0.f16)):
D0.f16 = S1.f16
elif GT_NEG_ZERO(S0.f16, S1.f16):
D0.f16 = S0.f16
else:
D0.f16 = S1.f16
# --- end pseudocode ---
def _SOP2Op_S_MUL_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = S0.f16 * S1.f16
# --- compiled pseudocode ---
D0.f16 = S0.f16 * S1.f16
# --- end pseudocode ---
def _SOP2Op_S_FMAC_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = fma(S0.f16, S1.f16, D0.f16)
# --- compiled pseudocode ---
D0.f16 = fma(S0.f16, S1.f16, D0.f16)
# --- end pseudocode ---
SOP2Op_FUNCTIONS = {
SOP2Op.S_ADD_U32: _SOP2Op_S_ADD_U32,
SOP2Op.S_SUB_U32: _SOP2Op_S_SUB_U32,
SOP2Op.S_ADD_I32: _SOP2Op_S_ADD_I32,
SOP2Op.S_SUB_I32: _SOP2Op_S_SUB_I32,
SOP2Op.S_ADDC_U32: _SOP2Op_S_ADDC_U32,
SOP2Op.S_SUBB_U32: _SOP2Op_S_SUBB_U32,
SOP2Op.S_ABSDIFF_I32: _SOP2Op_S_ABSDIFF_I32,
SOP2Op.S_LSHL_B32: _SOP2Op_S_LSHL_B32,
SOP2Op.S_LSHL_B64: _SOP2Op_S_LSHL_B64,
SOP2Op.S_LSHR_B32: _SOP2Op_S_LSHR_B32,
SOP2Op.S_LSHR_B64: _SOP2Op_S_LSHR_B64,
SOP2Op.S_ASHR_I32: _SOP2Op_S_ASHR_I32,
SOP2Op.S_ASHR_I64: _SOP2Op_S_ASHR_I64,
SOP2Op.S_LSHL1_ADD_U32: _SOP2Op_S_LSHL1_ADD_U32,
SOP2Op.S_LSHL2_ADD_U32: _SOP2Op_S_LSHL2_ADD_U32,
SOP2Op.S_LSHL3_ADD_U32: _SOP2Op_S_LSHL3_ADD_U32,
SOP2Op.S_LSHL4_ADD_U32: _SOP2Op_S_LSHL4_ADD_U32,
SOP2Op.S_MIN_I32: _SOP2Op_S_MIN_I32,
SOP2Op.S_MIN_U32: _SOP2Op_S_MIN_U32,
SOP2Op.S_MAX_I32: _SOP2Op_S_MAX_I32,
SOP2Op.S_MAX_U32: _SOP2Op_S_MAX_U32,
SOP2Op.S_AND_B32: _SOP2Op_S_AND_B32,
SOP2Op.S_AND_B64: _SOP2Op_S_AND_B64,
SOP2Op.S_OR_B32: _SOP2Op_S_OR_B32,
SOP2Op.S_OR_B64: _SOP2Op_S_OR_B64,
SOP2Op.S_XOR_B32: _SOP2Op_S_XOR_B32,
SOP2Op.S_XOR_B64: _SOP2Op_S_XOR_B64,
SOP2Op.S_NAND_B32: _SOP2Op_S_NAND_B32,
SOP2Op.S_NAND_B64: _SOP2Op_S_NAND_B64,
SOP2Op.S_NOR_B32: _SOP2Op_S_NOR_B32,
SOP2Op.S_NOR_B64: _SOP2Op_S_NOR_B64,
SOP2Op.S_XNOR_B32: _SOP2Op_S_XNOR_B32,
SOP2Op.S_XNOR_B64: _SOP2Op_S_XNOR_B64,
SOP2Op.S_AND_NOT1_B32: _SOP2Op_S_AND_NOT1_B32,
SOP2Op.S_AND_NOT1_B64: _SOP2Op_S_AND_NOT1_B64,
SOP2Op.S_OR_NOT1_B32: _SOP2Op_S_OR_NOT1_B32,
SOP2Op.S_OR_NOT1_B64: _SOP2Op_S_OR_NOT1_B64,
SOP2Op.S_BFE_U32: _SOP2Op_S_BFE_U32,
SOP2Op.S_BFE_I32: _SOP2Op_S_BFE_I32,
SOP2Op.S_BFE_U64: _SOP2Op_S_BFE_U64,
SOP2Op.S_BFE_I64: _SOP2Op_S_BFE_I64,
SOP2Op.S_BFM_B32: _SOP2Op_S_BFM_B32,
SOP2Op.S_BFM_B64: _SOP2Op_S_BFM_B64,
SOP2Op.S_MUL_I32: _SOP2Op_S_MUL_I32,
SOP2Op.S_MUL_HI_U32: _SOP2Op_S_MUL_HI_U32,
SOP2Op.S_MUL_HI_I32: _SOP2Op_S_MUL_HI_I32,
SOP2Op.S_CSELECT_B32: _SOP2Op_S_CSELECT_B32,
SOP2Op.S_CSELECT_B64: _SOP2Op_S_CSELECT_B64,
SOP2Op.S_PACK_LL_B32_B16: _SOP2Op_S_PACK_LL_B32_B16,
SOP2Op.S_PACK_LH_B32_B16: _SOP2Op_S_PACK_LH_B32_B16,
SOP2Op.S_PACK_HH_B32_B16: _SOP2Op_S_PACK_HH_B32_B16,
SOP2Op.S_PACK_HL_B32_B16: _SOP2Op_S_PACK_HL_B32_B16,
SOP2Op.S_ADD_F32: _SOP2Op_S_ADD_F32,
SOP2Op.S_SUB_F32: _SOP2Op_S_SUB_F32,
SOP2Op.S_MIN_F32: _SOP2Op_S_MIN_F32,
SOP2Op.S_MAX_F32: _SOP2Op_S_MAX_F32,
SOP2Op.S_MUL_F32: _SOP2Op_S_MUL_F32,
SOP2Op.S_FMAAK_F32: _SOP2Op_S_FMAAK_F32,
SOP2Op.S_FMAMK_F32: _SOP2Op_S_FMAMK_F32,
SOP2Op.S_FMAC_F32: _SOP2Op_S_FMAC_F32,
SOP2Op.S_CVT_PK_RTZ_F16_F32: _SOP2Op_S_CVT_PK_RTZ_F16_F32,
SOP2Op.S_ADD_F16: _SOP2Op_S_ADD_F16,
SOP2Op.S_SUB_F16: _SOP2Op_S_SUB_F16,
SOP2Op.S_MIN_F16: _SOP2Op_S_MIN_F16,
SOP2Op.S_MAX_F16: _SOP2Op_S_MAX_F16,
SOP2Op.S_MUL_F16: _SOP2Op_S_MUL_F16,
SOP2Op.S_FMAC_F16: _SOP2Op_S_FMAC_F16,
}
def _SOPCOp_S_CMP_EQ_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.i32 == S1.i32
# --- compiled pseudocode ---
SCC.b32 = S0.i32 == S1.i32
# --- end pseudocode ---
def _SOPCOp_S_CMP_LG_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.i32 <> S1.i32
# --- compiled pseudocode ---
SCC.b32 = S0.i32 != S1.i32
# --- end pseudocode ---
def _SOPCOp_S_CMP_GT_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.i32 > S1.i32
# --- compiled pseudocode ---
SCC.b32 = S0.i32 > S1.i32
# --- end pseudocode ---
def _SOPCOp_S_CMP_GE_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.i32 >= S1.i32
# --- compiled pseudocode ---
SCC.b32 = S0.i32 >= S1.i32
# --- end pseudocode ---
def _SOPCOp_S_CMP_LT_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.i32 < S1.i32
# --- compiled pseudocode ---
SCC.b32 = S0.i32 < S1.i32
# --- end pseudocode ---
def _SOPCOp_S_CMP_LE_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.i32 <= S1.i32
# --- compiled pseudocode ---
SCC.b32 = S0.i32 <= S1.i32
# --- end pseudocode ---
def _SOPCOp_S_CMP_EQ_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.u32 == S1.u32
# --- compiled pseudocode ---
SCC.b32 = S0.u32 == S1.u32
# --- end pseudocode ---
def _SOPCOp_S_CMP_LG_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.u32 <> S1.u32
# --- compiled pseudocode ---
SCC.b32 = S0.u32 != S1.u32
# --- end pseudocode ---
def _SOPCOp_S_CMP_GT_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.u32 > S1.u32
# --- compiled pseudocode ---
SCC.b32 = S0.u32 > S1.u32
# --- end pseudocode ---
def _SOPCOp_S_CMP_GE_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.u32 >= S1.u32
# --- compiled pseudocode ---
SCC.b32 = S0.u32 >= S1.u32
# --- end pseudocode ---
def _SOPCOp_S_CMP_LT_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.u32 < S1.u32
# --- compiled pseudocode ---
SCC.b32 = S0.u32 < S1.u32
# --- end pseudocode ---
def _SOPCOp_S_CMP_LE_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.u32 <= S1.u32
# --- compiled pseudocode ---
SCC.b32 = S0.u32 <= S1.u32
# --- end pseudocode ---
def _SOPCOp_S_BITCMP0_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.u32[S1.u32[4 : 0]] == 1'0U
# --- compiled pseudocode ---
SCC.b32 = S0.u32[S1.u32[4 : 0]] == 0
# --- end pseudocode ---
def _SOPCOp_S_BITCMP1_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.u32[S1.u32[4 : 0]] == 1'1U
# --- compiled pseudocode ---
SCC.b32 = S0.u32[S1.u32[4 : 0]] == 1
# --- end pseudocode ---
def _SOPCOp_S_BITCMP0_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.u64[S1.u32[5 : 0]] == 1'0U
# --- compiled pseudocode ---
SCC.b32 = S0.u64[S1.u32[5 : 0]] == 0
# --- end pseudocode ---
def _SOPCOp_S_BITCMP1_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.u64[S1.u32[5 : 0]] == 1'1U
# --- compiled pseudocode ---
SCC.b32 = S0.u64[S1.u32[5 : 0]] == 1
# --- end pseudocode ---
def _SOPCOp_S_CMP_EQ_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.u64 == S1.u64
# --- compiled pseudocode ---
SCC.b32 = S0.u64 == S1.u64
# --- end pseudocode ---
def _SOPCOp_S_CMP_LG_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.u64 <> S1.u64
# --- compiled pseudocode ---
SCC.b32 = S0.u64 != S1.u64
# --- end pseudocode ---
def _SOPCOp_S_CMP_LT_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.f32 < S1.f32
# --- compiled pseudocode ---
SCC.b32 = S0.f32 < S1.f32
# --- end pseudocode ---
def _SOPCOp_S_CMP_LT_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.f16 < S1.f16
# --- compiled pseudocode ---
SCC.b32 = S0.f16 < S1.f16
# --- end pseudocode ---
def _SOPCOp_S_CMP_EQ_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.f32 == S1.f32
# --- compiled pseudocode ---
SCC.b32 = S0.f32 == S1.f32
# --- end pseudocode ---
def _SOPCOp_S_CMP_EQ_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.f16 == S1.f16
# --- compiled pseudocode ---
SCC.b32 = S0.f16 == S1.f16
# --- end pseudocode ---
def _SOPCOp_S_CMP_LE_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.f32 <= S1.f32
# --- compiled pseudocode ---
SCC.b32 = S0.f32 <= S1.f32
# --- end pseudocode ---
def _SOPCOp_S_CMP_LE_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.f16 <= S1.f16
# --- compiled pseudocode ---
SCC.b32 = S0.f16 <= S1.f16
# --- end pseudocode ---
def _SOPCOp_S_CMP_GT_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.f32 > S1.f32
# --- compiled pseudocode ---
SCC.b32 = S0.f32 > S1.f32
# --- end pseudocode ---
def _SOPCOp_S_CMP_GT_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.f16 > S1.f16
# --- compiled pseudocode ---
SCC.b32 = S0.f16 > S1.f16
# --- end pseudocode ---
def _SOPCOp_S_CMP_LG_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.f32 <> S1.f32
# --- compiled pseudocode ---
SCC.b32 = S0.f32 != S1.f32
# --- end pseudocode ---
def _SOPCOp_S_CMP_LG_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.f16 <> S1.f16
# --- compiled pseudocode ---
SCC.b32 = S0.f16 != S1.f16
# --- end pseudocode ---
def _SOPCOp_S_CMP_GE_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.f32 >= S1.f32
# --- compiled pseudocode ---
SCC.b32 = S0.f32 >= S1.f32
# --- end pseudocode ---
def _SOPCOp_S_CMP_GE_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.f16 >= S1.f16
# --- compiled pseudocode ---
SCC.b32 = S0.f16 >= S1.f16
# --- end pseudocode ---
def _SOPCOp_S_CMP_O_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = (!isNAN(64'F(S0.f32)) && !isNAN(64'F(S1.f32)))
# --- compiled pseudocode ---
SCC.b32 = ( not isNAN(F(S0.f32)) and not isNAN(F(S1.f32)))
# --- end pseudocode ---
def _SOPCOp_S_CMP_O_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = (!isNAN(64'F(S0.f16)) && !isNAN(64'F(S1.f16)))
# --- compiled pseudocode ---
SCC.b32 = ( not isNAN(F(S0.f16)) and not isNAN(F(S1.f16)))
# --- end pseudocode ---
def _SOPCOp_S_CMP_U_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = (isNAN(64'F(S0.f32)) || isNAN(64'F(S1.f32)))
# --- compiled pseudocode ---
SCC.b32 = (isNAN(F(S0.f32)) or isNAN(F(S1.f32)))
# --- end pseudocode ---
def _SOPCOp_S_CMP_U_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = (isNAN(64'F(S0.f16)) || isNAN(64'F(S1.f16)))
# --- compiled pseudocode ---
SCC.b32 = (isNAN(F(S0.f16)) or isNAN(F(S1.f16)))
# --- end pseudocode ---
def _SOPCOp_S_CMP_NGE_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = !(S0.f32 >= S1.f32);
# // With NAN inputs this is not the same operation as <
# --- compiled pseudocode ---
SCC.b32 = not (S0.f32 >= S1.f32)
# --- end pseudocode ---
def _SOPCOp_S_CMP_NGE_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = !(S0.f16 >= S1.f16);
# // With NAN inputs this is not the same operation as <
# --- compiled pseudocode ---
SCC.b32 = not (S0.f16 >= S1.f16)
# --- end pseudocode ---
def _SOPCOp_S_CMP_NLG_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = !(S0.f32 <> S1.f32);
# // With NAN inputs this is not the same operation as ==
# --- compiled pseudocode ---
SCC.b32 = not (S0.f32 != S1.f32)
# --- end pseudocode ---
def _SOPCOp_S_CMP_NLG_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = !(S0.f16 <> S1.f16);
# // With NAN inputs this is not the same operation as ==
# --- compiled pseudocode ---
SCC.b32 = not (S0.f16 != S1.f16)
# --- end pseudocode ---
def _SOPCOp_S_CMP_NGT_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = !(S0.f32 > S1.f32);
# // With NAN inputs this is not the same operation as <=
# --- compiled pseudocode ---
SCC.b32 = not (S0.f32 > S1.f32)
# --- end pseudocode ---
def _SOPCOp_S_CMP_NGT_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = !(S0.f16 > S1.f16);
# // With NAN inputs this is not the same operation as <=
# --- compiled pseudocode ---
SCC.b32 = not (S0.f16 > S1.f16)
# --- end pseudocode ---
def _SOPCOp_S_CMP_NLE_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = !(S0.f32 <= S1.f32);
# // With NAN inputs this is not the same operation as >
# --- compiled pseudocode ---
SCC.b32 = not (S0.f32 <= S1.f32)
# --- end pseudocode ---
def _SOPCOp_S_CMP_NLE_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = !(S0.f16 <= S1.f16);
# // With NAN inputs this is not the same operation as >
# --- compiled pseudocode ---
SCC.b32 = not (S0.f16 <= S1.f16)
# --- end pseudocode ---
def _SOPCOp_S_CMP_NEQ_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = !(S0.f32 == S1.f32);
# // With NAN inputs this is not the same operation as !=
# --- compiled pseudocode ---
SCC.b32 = not (S0.f32 == S1.f32)
# --- end pseudocode ---
def _SOPCOp_S_CMP_NEQ_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = !(S0.f16 == S1.f16);
# // With NAN inputs this is not the same operation as !=
# --- compiled pseudocode ---
SCC.b32 = not (S0.f16 == S1.f16)
# --- end pseudocode ---
def _SOPCOp_S_CMP_NLT_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = !(S0.f32 < S1.f32);
# // With NAN inputs this is not the same operation as >=
# --- compiled pseudocode ---
SCC.b32 = not (S0.f32 < S1.f32)
# --- end pseudocode ---
def _SOPCOp_S_CMP_NLT_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = !(S0.f16 < S1.f16);
# // With NAN inputs this is not the same operation as >=
# --- compiled pseudocode ---
SCC.b32 = not (S0.f16 < S1.f16)
# --- end pseudocode ---
SOPCOp_FUNCTIONS = {
SOPCOp.S_CMP_EQ_I32: _SOPCOp_S_CMP_EQ_I32,
SOPCOp.S_CMP_LG_I32: _SOPCOp_S_CMP_LG_I32,
SOPCOp.S_CMP_GT_I32: _SOPCOp_S_CMP_GT_I32,
SOPCOp.S_CMP_GE_I32: _SOPCOp_S_CMP_GE_I32,
SOPCOp.S_CMP_LT_I32: _SOPCOp_S_CMP_LT_I32,
SOPCOp.S_CMP_LE_I32: _SOPCOp_S_CMP_LE_I32,
SOPCOp.S_CMP_EQ_U32: _SOPCOp_S_CMP_EQ_U32,
SOPCOp.S_CMP_LG_U32: _SOPCOp_S_CMP_LG_U32,
SOPCOp.S_CMP_GT_U32: _SOPCOp_S_CMP_GT_U32,
SOPCOp.S_CMP_GE_U32: _SOPCOp_S_CMP_GE_U32,
SOPCOp.S_CMP_LT_U32: _SOPCOp_S_CMP_LT_U32,
SOPCOp.S_CMP_LE_U32: _SOPCOp_S_CMP_LE_U32,
SOPCOp.S_BITCMP0_B32: _SOPCOp_S_BITCMP0_B32,
SOPCOp.S_BITCMP1_B32: _SOPCOp_S_BITCMP1_B32,
SOPCOp.S_BITCMP0_B64: _SOPCOp_S_BITCMP0_B64,
SOPCOp.S_BITCMP1_B64: _SOPCOp_S_BITCMP1_B64,
SOPCOp.S_CMP_EQ_U64: _SOPCOp_S_CMP_EQ_U64,
SOPCOp.S_CMP_LG_U64: _SOPCOp_S_CMP_LG_U64,
SOPCOp.S_CMP_LT_F32: _SOPCOp_S_CMP_LT_F32,
SOPCOp.S_CMP_LT_F16: _SOPCOp_S_CMP_LT_F16,
SOPCOp.S_CMP_EQ_F32: _SOPCOp_S_CMP_EQ_F32,
SOPCOp.S_CMP_EQ_F16: _SOPCOp_S_CMP_EQ_F16,
SOPCOp.S_CMP_LE_F32: _SOPCOp_S_CMP_LE_F32,
SOPCOp.S_CMP_LE_F16: _SOPCOp_S_CMP_LE_F16,
SOPCOp.S_CMP_GT_F32: _SOPCOp_S_CMP_GT_F32,
SOPCOp.S_CMP_GT_F16: _SOPCOp_S_CMP_GT_F16,
SOPCOp.S_CMP_LG_F32: _SOPCOp_S_CMP_LG_F32,
SOPCOp.S_CMP_LG_F16: _SOPCOp_S_CMP_LG_F16,
SOPCOp.S_CMP_GE_F32: _SOPCOp_S_CMP_GE_F32,
SOPCOp.S_CMP_GE_F16: _SOPCOp_S_CMP_GE_F16,
SOPCOp.S_CMP_O_F32: _SOPCOp_S_CMP_O_F32,
SOPCOp.S_CMP_O_F16: _SOPCOp_S_CMP_O_F16,
SOPCOp.S_CMP_U_F32: _SOPCOp_S_CMP_U_F32,
SOPCOp.S_CMP_U_F16: _SOPCOp_S_CMP_U_F16,
SOPCOp.S_CMP_NGE_F32: _SOPCOp_S_CMP_NGE_F32,
SOPCOp.S_CMP_NGE_F16: _SOPCOp_S_CMP_NGE_F16,
SOPCOp.S_CMP_NLG_F32: _SOPCOp_S_CMP_NLG_F32,
SOPCOp.S_CMP_NLG_F16: _SOPCOp_S_CMP_NLG_F16,
SOPCOp.S_CMP_NGT_F32: _SOPCOp_S_CMP_NGT_F32,
SOPCOp.S_CMP_NGT_F16: _SOPCOp_S_CMP_NGT_F16,
SOPCOp.S_CMP_NLE_F32: _SOPCOp_S_CMP_NLE_F32,
SOPCOp.S_CMP_NLE_F16: _SOPCOp_S_CMP_NLE_F16,
SOPCOp.S_CMP_NEQ_F32: _SOPCOp_S_CMP_NEQ_F32,
SOPCOp.S_CMP_NEQ_F16: _SOPCOp_S_CMP_NEQ_F16,
SOPCOp.S_CMP_NLT_F32: _SOPCOp_S_CMP_NLT_F32,
SOPCOp.S_CMP_NLT_F16: _SOPCOp_S_CMP_NLT_F16,
}
def _SOPKOp_S_MOVK_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = 32'I(signext(SIMM16.i16))
# --- compiled pseudocode ---
D0.i32 = (signext(SIMM16.i16))
# --- end pseudocode ---
def _SOPKOp_S_VERSION(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# // Do nothing - for use by tools only
# --- compiled pseudocode ---
# --- end pseudocode ---
pass
def _SOPKOp_S_CMOVK_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if SCC then
# D0.i32 = 32'I(signext(SIMM16.i16))
# endif
# --- compiled pseudocode ---
if SCC:
D0.i32 = (signext(SIMM16.i16))
# --- end pseudocode ---
def _SOPKOp_S_CMPK_EQ_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = 64'I(S0.i32) == signext(SIMM16.i16)
# --- compiled pseudocode ---
SCC.b32 = (S0.i32) == signext(SIMM16.i16)
# --- end pseudocode ---
def _SOPKOp_S_CMPK_LG_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = 64'I(S0.i32) != signext(SIMM16.i16)
# --- compiled pseudocode ---
SCC.b32 = (S0.i32) != signext(SIMM16.i16)
# --- end pseudocode ---
def _SOPKOp_S_CMPK_GT_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = 64'I(S0.i32) > signext(SIMM16.i16)
# --- compiled pseudocode ---
SCC.b32 = (S0.i32) > signext(SIMM16.i16)
# --- end pseudocode ---
def _SOPKOp_S_CMPK_GE_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = 64'I(S0.i32) >= signext(SIMM16.i16)
# --- compiled pseudocode ---
SCC.b32 = (S0.i32) >= signext(SIMM16.i16)
# --- end pseudocode ---
def _SOPKOp_S_CMPK_LT_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = 64'I(S0.i32) < signext(SIMM16.i16)
# --- compiled pseudocode ---
SCC.b32 = (S0.i32) < signext(SIMM16.i16)
# --- end pseudocode ---
def _SOPKOp_S_CMPK_LE_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = 64'I(S0.i32) <= signext(SIMM16.i16)
# --- compiled pseudocode ---
SCC.b32 = (S0.i32) <= signext(SIMM16.i16)
# --- end pseudocode ---
def _SOPKOp_S_CMPK_EQ_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.u32 == 32'U(SIMM16.u16)
# --- compiled pseudocode ---
SCC.b32 = S0.u32 == (SIMM16.u16)
# --- end pseudocode ---
def _SOPKOp_S_CMPK_LG_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.u32 != 32'U(SIMM16.u16)
# --- compiled pseudocode ---
SCC.b32 = S0.u32 != (SIMM16.u16)
# --- end pseudocode ---
def _SOPKOp_S_CMPK_GT_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.u32 > 32'U(SIMM16.u16)
# --- compiled pseudocode ---
SCC.b32 = S0.u32 > (SIMM16.u16)
# --- end pseudocode ---
def _SOPKOp_S_CMPK_GE_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.u32 >= 32'U(SIMM16.u16)
# --- compiled pseudocode ---
SCC.b32 = S0.u32 >= (SIMM16.u16)
# --- end pseudocode ---
def _SOPKOp_S_CMPK_LT_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.u32 < 32'U(SIMM16.u16)
# --- compiled pseudocode ---
SCC.b32 = S0.u32 < (SIMM16.u16)
# --- end pseudocode ---
def _SOPKOp_S_CMPK_LE_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# SCC = S0.u32 <= 32'U(SIMM16.u16)
# --- compiled pseudocode ---
SCC.b32 = S0.u32 <= (SIMM16.u16)
# --- end pseudocode ---
def _SOPKOp_S_ADDK_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = D0.i32;
# D0.i32 = 32'I(64'I(D0.i32) + signext(SIMM16.i16));
# SCC = ((tmp[31] == SIMM16.i16[15]) && (tmp[31] != D0.i32[31]));
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(D0.i32)
D0.i32 = ((D0.i32) + signext(SIMM16.i16))
SCC.b32 = ((tmp[31] == SIMM16.i16[15]) and (tmp[31] != D0.i32[31]))
# --- end pseudocode ---
def _SOPKOp_S_MULK_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = 32'I(64'I(D0.i32) * signext(SIMM16.i16))
# --- compiled pseudocode ---
D0.i32 = ((D0.i32) * signext(SIMM16.i16))
# --- end pseudocode ---
SOPKOp_FUNCTIONS = {
SOPKOp.S_MOVK_I32: _SOPKOp_S_MOVK_I32,
SOPKOp.S_VERSION: _SOPKOp_S_VERSION,
SOPKOp.S_CMOVK_I32: _SOPKOp_S_CMOVK_I32,
SOPKOp.S_CMPK_EQ_I32: _SOPKOp_S_CMPK_EQ_I32,
SOPKOp.S_CMPK_LG_I32: _SOPKOp_S_CMPK_LG_I32,
SOPKOp.S_CMPK_GT_I32: _SOPKOp_S_CMPK_GT_I32,
SOPKOp.S_CMPK_GE_I32: _SOPKOp_S_CMPK_GE_I32,
SOPKOp.S_CMPK_LT_I32: _SOPKOp_S_CMPK_LT_I32,
SOPKOp.S_CMPK_LE_I32: _SOPKOp_S_CMPK_LE_I32,
SOPKOp.S_CMPK_EQ_U32: _SOPKOp_S_CMPK_EQ_U32,
SOPKOp.S_CMPK_LG_U32: _SOPKOp_S_CMPK_LG_U32,
SOPKOp.S_CMPK_GT_U32: _SOPKOp_S_CMPK_GT_U32,
SOPKOp.S_CMPK_GE_U32: _SOPKOp_S_CMPK_GE_U32,
SOPKOp.S_CMPK_LT_U32: _SOPKOp_S_CMPK_LT_U32,
SOPKOp.S_CMPK_LE_U32: _SOPKOp_S_CMPK_LE_U32,
SOPKOp.S_ADDK_I32: _SOPKOp_S_ADDK_I32,
SOPKOp.S_MULK_I32: _SOPKOp_S_MULK_I32,
}
def _SOPPOp_S_NOP(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# for i in 0U : SIMM16.u16[3 : 0].u32 do
# endfor
# --- compiled pseudocode ---
for i in range(0, int(SIMM16.u16[3 : 0].u32)+1):
pass
# --- end pseudocode ---
def _SOPPOp_S_DELAY_ALU(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# instruction may be omitted. For wave64 the compiler may not know the status of the EXEC mask and hence
# // 1 cycle delay here
# // 2 cycles delay here
# --- compiled pseudocode ---
# --- end pseudocode ---
pass
def _SOPPOp_S_TRAP(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# // PC passed into trap handler points to S_TRAP itself,
# // trap base address
# --- compiled pseudocode ---
# --- end pseudocode ---
pass
SOPPOp_FUNCTIONS = {
SOPPOp.S_NOP: _SOPPOp_S_NOP,
SOPPOp.S_DELAY_ALU: _SOPPOp_S_DELAY_ALU,
SOPPOp.S_TRAP: _SOPPOp_S_TRAP,
}
def _VOP1Op_V_MOV_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.b32 = S0.b32
# --- compiled pseudocode ---
D0.b32 = S0.b32
# --- end pseudocode ---
def _VOP1Op_V_READFIRSTLANE_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# declare lane : 32'U;
# if WAVE64 then
# // 64 lanes
# if EXEC == 0x0LL then
# lane = 0U;
# // Force lane 0 if all lanes are disabled
# else
# lane = 32'U(s_ff1_i32_b64(EXEC));
# // Lowest active lane
# endif
# else
# // 32 lanes
# if EXEC_LO.i32 == 0 then
# lane = 0U;
# // Force lane 0 if all lanes are disabled
# else
# lane = 32'U(s_ff1_i32_b32(EXEC_LO));
# // Lowest active lane
# endif
# endif;
# D0.b32 = VGPR[lane][SRC0.u32]
EXEC_LO = SliceProxy(EXEC, 31, 0)
# --- compiled pseudocode ---
if WAVE64:
if EXEC == 0x0:
lane = 0
else:
lane = (s_ff1_i32_b64(EXEC))
else:
if EXEC_LO.i32 == 0:
lane = 0
else:
lane = (s_ff1_i32_b32(EXEC_LO))
D0.b32 = VGPR[lane][SRC0.u32]
# --- end pseudocode ---
def _VOP1Op_V_CVT_I32_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = f64_to_i32(S0.f64)
# --- compiled pseudocode ---
D0.i32 = f64_to_i32(S0.f64)
# --- end pseudocode ---
def _VOP1Op_V_CVT_F64_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f64 = i32_to_f64(S0.i32)
# --- compiled pseudocode ---
D0.f64 = i32_to_f64(S0.i32)
# --- end pseudocode ---
def _VOP1Op_V_CVT_F32_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = i32_to_f32(S0.i32)
# --- compiled pseudocode ---
D0.f32 = i32_to_f32(S0.i32)
# --- end pseudocode ---
def _VOP1Op_V_CVT_F32_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = u32_to_f32(S0.u32)
# --- compiled pseudocode ---
D0.f32 = u32_to_f32(S0.u32)
# --- end pseudocode ---
def _VOP1Op_V_CVT_U32_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = f32_to_u32(S0.f32)
# --- compiled pseudocode ---
D0.u32 = f32_to_u32(S0.f32)
# --- end pseudocode ---
def _VOP1Op_V_CVT_I32_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = f32_to_i32(S0.f32)
# --- compiled pseudocode ---
D0.i32 = f32_to_i32(S0.f32)
# --- end pseudocode ---
def _VOP1Op_V_CVT_F16_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = f32_to_f16(S0.f32)
# --- compiled pseudocode ---
D0.f16 = f32_to_f16(S0.f32)
# --- end pseudocode ---
def _VOP1Op_V_CVT_F32_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = f16_to_f32(S0.f16)
# --- compiled pseudocode ---
D0.f32 = f16_to_f32(S0.f16)
# --- end pseudocode ---
def _VOP1Op_V_CVT_NEAREST_I32_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = f32_to_i32(floor(S0.f32 + 0.5F))
# --- compiled pseudocode ---
D0.i32 = f32_to_i32(floor(S0.f32 + 0.5))
# --- end pseudocode ---
def _VOP1Op_V_CVT_FLOOR_I32_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = f32_to_i32(floor(S0.f32))
# --- compiled pseudocode ---
D0.i32 = f32_to_i32(floor(S0.f32))
# --- end pseudocode ---
def _VOP1Op_V_CVT_F32_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = f64_to_f32(S0.f64)
# --- compiled pseudocode ---
D0.f32 = f64_to_f32(S0.f64)
# --- end pseudocode ---
def _VOP1Op_V_CVT_F64_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f64 = f32_to_f64(S0.f32)
# --- compiled pseudocode ---
D0.f64 = f32_to_f64(S0.f32)
# --- end pseudocode ---
def _VOP1Op_V_CVT_F32_UBYTE0(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = u32_to_f32(S0[7 : 0].u32)
# --- compiled pseudocode ---
D0.f32 = u32_to_f32(S0[7 : 0].u32)
# --- end pseudocode ---
def _VOP1Op_V_CVT_F32_UBYTE1(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = u32_to_f32(S0[15 : 8].u32)
# --- compiled pseudocode ---
D0.f32 = u32_to_f32(S0[15 : 8].u32)
# --- end pseudocode ---
def _VOP1Op_V_CVT_F32_UBYTE2(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = u32_to_f32(S0[23 : 16].u32)
# --- compiled pseudocode ---
D0.f32 = u32_to_f32(S0[23 : 16].u32)
# --- end pseudocode ---
def _VOP1Op_V_CVT_F32_UBYTE3(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = u32_to_f32(S0[31 : 24].u32)
# --- compiled pseudocode ---
D0.f32 = u32_to_f32(S0[31 : 24].u32)
# --- end pseudocode ---
def _VOP1Op_V_CVT_U32_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = f64_to_u32(S0.f64)
# --- compiled pseudocode ---
D0.u32 = f64_to_u32(S0.f64)
# --- end pseudocode ---
def _VOP1Op_V_CVT_F64_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f64 = u32_to_f64(S0.u32)
# --- compiled pseudocode ---
D0.f64 = u32_to_f64(S0.u32)
# --- end pseudocode ---
def _VOP1Op_V_TRUNC_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f64 = trunc(S0.f64)
# --- compiled pseudocode ---
D0.f64 = trunc(S0.f64)
# --- end pseudocode ---
def _VOP1Op_V_CEIL_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f64 = trunc(S0.f64);
# if ((S0.f64 > 0.0) && (S0.f64 != D0.f64)) then
# D0.f64 += 1.0
# endif
# --- compiled pseudocode ---
D0.f64 = trunc(S0.f64)
if ((S0.f64 > 0.0) and (S0.f64 != D0.f64)):
D0.f64 += 1.0
# --- end pseudocode ---
def _VOP1Op_V_RNDNE_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f64 = floor(S0.f64 + 0.5);
# if (isEven(floor(S0.f64)) && (fract(S0.f64) == 0.5)) then
# D0.f64 -= 1.0
# endif
# --- compiled pseudocode ---
D0.f64 = floor(S0.f64 + 0.5)
if (isEven(floor(S0.f64)) and (fract(S0.f64) == 0.5)):
D0.f64 -= 1.0
# --- end pseudocode ---
def _VOP1Op_V_FLOOR_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f64 = trunc(S0.f64);
# if ((S0.f64 < 0.0) && (S0.f64 != D0.f64)) then
# D0.f64 += -1.0
# endif
# --- compiled pseudocode ---
D0.f64 = trunc(S0.f64)
if ((S0.f64 < 0.0) and (S0.f64 != D0.f64)):
D0.f64 += -1.0
# --- end pseudocode ---
def _VOP1Op_V_MOV_B16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.b16 = S0.b16
# --- compiled pseudocode ---
D0.b16 = S0.b16
# --- end pseudocode ---
def _VOP1Op_V_FRACT_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = S0.f32 + -floor(S0.f32)
# --- compiled pseudocode ---
D0.f32 = S0.f32 + -floor(S0.f32)
# --- end pseudocode ---
def _VOP1Op_V_TRUNC_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = trunc(S0.f32)
# --- compiled pseudocode ---
D0.f32 = trunc(S0.f32)
# --- end pseudocode ---
def _VOP1Op_V_CEIL_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = trunc(S0.f32);
# if ((S0.f32 > 0.0F) && (S0.f32 != D0.f32)) then
# D0.f32 += 1.0F
# endif
# --- compiled pseudocode ---
D0.f32 = trunc(S0.f32)
if ((S0.f32 > 0.0) and (S0.f32 != D0.f32)):
D0.f32 += 1.0
# --- end pseudocode ---
def _VOP1Op_V_RNDNE_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = floor(S0.f32 + 0.5F);
# if (isEven(64'F(floor(S0.f32))) && (fract(S0.f32) == 0.5F)) then
# D0.f32 -= 1.0F
# endif
# --- compiled pseudocode ---
D0.f32 = floor(S0.f32 + 0.5)
if (isEven(F(floor(S0.f32))) and (fract(S0.f32) == 0.5)):
D0.f32 -= 1.0
# --- end pseudocode ---
def _VOP1Op_V_FLOOR_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = trunc(S0.f32);
# if ((S0.f32 < 0.0F) && (S0.f32 != D0.f32)) then
# D0.f32 += -1.0F
# endif
# --- compiled pseudocode ---
D0.f32 = trunc(S0.f32)
if ((S0.f32 < 0.0) and (S0.f32 != D0.f32)):
D0.f32 += -1.0
# --- end pseudocode ---
def _VOP1Op_V_EXP_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = pow(2.0F, S0.f32)
# --- compiled pseudocode ---
D0.f32 = pow(2.0, S0.f32)
# --- end pseudocode ---
def _VOP1Op_V_LOG_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = log2(S0.f32)
# --- compiled pseudocode ---
D0.f32 = log2(S0.f32)
# --- end pseudocode ---
def _VOP1Op_V_RCP_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = 1.0F / S0.f32
# --- compiled pseudocode ---
D0.f32 = 1.0 / S0.f32
# --- end pseudocode ---
def _VOP1Op_V_RCP_IFLAG_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = 1.0F / S0.f32;
# // Can only raise integer DIV_BY_ZERO exception
# --- compiled pseudocode ---
D0.f32 = 1.0 / S0.f32
# --- end pseudocode ---
def _VOP1Op_V_RSQ_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = 1.0F / sqrt(S0.f32)
# --- compiled pseudocode ---
D0.f32 = 1.0 / sqrt(S0.f32)
# --- end pseudocode ---
def _VOP1Op_V_RCP_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f64 = 1.0 / S0.f64
# --- compiled pseudocode ---
D0.f64 = 1.0 / S0.f64
# --- end pseudocode ---
def _VOP1Op_V_RSQ_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f64 = 1.0 / sqrt(S0.f64)
# --- compiled pseudocode ---
D0.f64 = 1.0 / sqrt(S0.f64)
# --- end pseudocode ---
def _VOP1Op_V_SQRT_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = sqrt(S0.f32)
# --- compiled pseudocode ---
D0.f32 = sqrt(S0.f32)
# --- end pseudocode ---
def _VOP1Op_V_SQRT_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f64 = sqrt(S0.f64)
# --- compiled pseudocode ---
D0.f64 = sqrt(S0.f64)
# --- end pseudocode ---
def _VOP1Op_V_SIN_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = sin(S0.f32 * 32'F(PI * 2.0))
# --- compiled pseudocode ---
D0.f32 = sin(S0.f32 * F(PI * 2.0))
# --- end pseudocode ---
def _VOP1Op_V_COS_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = cos(S0.f32 * 32'F(PI * 2.0))
# --- compiled pseudocode ---
D0.f32 = cos(S0.f32 * F(PI * 2.0))
# --- end pseudocode ---
def _VOP1Op_V_NOT_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = ~S0.u32
# --- compiled pseudocode ---
D0.u32 = ~S0.u32
# --- end pseudocode ---
def _VOP1Op_V_BFREV_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32[31 : 0] = S0.u32[0 : 31]
# --- compiled pseudocode ---
D0.u32[31 : 0] = S0.u32[0 : 31]
# --- end pseudocode ---
def _VOP1Op_V_CLZ_I32_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = -1;
# // Set if no ones are found
# for i in 0 : 31 do
# // Search from MSB
# if S0.u32[31 - i] == 1'1U then
# D0.i32 = i;
# endif
# endfor
# --- compiled pseudocode ---
D0.i32 = -1
for i in range(0, int(31)+1):
if S0.u32[31 - i] == 1:
D0.i32 = i; break
# --- end pseudocode ---
def _VOP1Op_V_CTZ_I32_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = -1;
# // Set if no ones are found
# for i in 0 : 31 do
# // Search from LSB
# if S0.u32[i] == 1'1U then
# D0.i32 = i;
# endif
# endfor
# --- compiled pseudocode ---
D0.i32 = -1
for i in range(0, int(31)+1):
if S0.u32[i] == 1:
D0.i32 = i; break
# --- end pseudocode ---
def _VOP1Op_V_CLS_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = -1;
# // Set if all bits are the same
# for i in 1 : 31 do
# // Search from MSB
# if S0.i32[31 - i] != S0.i32[31] then
# D0.i32 = i;
# endif
# endfor
# --- compiled pseudocode ---
D0.i32 = -1
for i in range(1, int(31)+1):
if S0.i32[31 - i] != S0.i32[31]:
D0.i32 = i
# --- end pseudocode ---
def _VOP1Op_V_FREXP_EXP_I32_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if ((S0.f64 == +INF) || (S0.f64 == -INF) || isNAN(S0.f64)) then
# D0.i32 = 0
# else
# D0.i32 = exponent(S0.f64) - 1023 + 1
# endif
# --- compiled pseudocode ---
if ((S0.f64 == INF) or (S0.f64 == (-INF)) or isNAN(S0.f64)):
D0.i32 = 0
else:
D0.i32 = exponent(S0.f64) - 1023 + 1
# --- end pseudocode ---
def _VOP1Op_V_FREXP_MANT_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if ((S0.f64 == +INF) || (S0.f64 == -INF) || isNAN(S0.f64)) then
# D0.f64 = S0.f64
# else
# D0.f64 = mantissa(S0.f64)
# endif
# --- compiled pseudocode ---
if ((S0.f64 == INF) or (S0.f64 == (-INF)) or isNAN(S0.f64)):
D0.f64 = S0.f64
else:
D0.f64 = mantissa(S0.f64)
# --- end pseudocode ---
def _VOP1Op_V_FRACT_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f64 = S0.f64 + -floor(S0.f64)
# --- compiled pseudocode ---
D0.f64 = S0.f64 + -floor(S0.f64)
# --- end pseudocode ---
def _VOP1Op_V_FREXP_EXP_I32_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if ((64'F(S0.f32) == +INF) || (64'F(S0.f32) == -INF) || isNAN(64'F(S0.f32))) then
# D0.i32 = 0
# else
# D0.i32 = exponent(S0.f32) - 127 + 1
# endif
# --- compiled pseudocode ---
if ((F(S0.f32) == INF) or (F(S0.f32) == (-INF)) or isNAN(F(S0.f32))):
D0.i32 = 0
else:
D0.i32 = exponent(S0.f32) - 127 + 1
# --- end pseudocode ---
def _VOP1Op_V_FREXP_MANT_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if ((64'F(S0.f32) == +INF) || (64'F(S0.f32) == -INF) || isNAN(64'F(S0.f32))) then
# D0.f32 = S0.f32
# else
# D0.f32 = mantissa(S0.f32)
# endif
# --- compiled pseudocode ---
if ((F(S0.f32) == INF) or (F(S0.f32) == (-INF)) or isNAN(F(S0.f32))):
D0.f32 = S0.f32
else:
D0.f32 = mantissa(S0.f32)
# --- end pseudocode ---
def _VOP1Op_V_MOVRELS_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# addr = SRC0.u32;
# // Raw value from instruction
# D0.b32 = VGPR[laneId][addr].b32
# --- compiled pseudocode ---
addr = SRC0.u32
D0.b32 = VGPR[laneId][addr].b32
# --- end pseudocode ---
def _VOP1Op_V_CVT_F16_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = u16_to_f16(S0.u16)
# --- compiled pseudocode ---
D0.f16 = u16_to_f16(S0.u16)
# --- end pseudocode ---
def _VOP1Op_V_CVT_F16_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = i16_to_f16(S0.i16)
# --- compiled pseudocode ---
D0.f16 = i16_to_f16(S0.i16)
# --- end pseudocode ---
def _VOP1Op_V_CVT_U16_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u16 = f16_to_u16(S0.f16)
# --- compiled pseudocode ---
D0.u16 = f16_to_u16(S0.f16)
# --- end pseudocode ---
def _VOP1Op_V_CVT_I16_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i16 = f16_to_i16(S0.f16)
# --- compiled pseudocode ---
D0.i16 = f16_to_i16(S0.f16)
# --- end pseudocode ---
def _VOP1Op_V_RCP_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = 16'1.0 / S0.f16
# --- compiled pseudocode ---
D0.f16 = 1.0 / S0.f16
# --- end pseudocode ---
def _VOP1Op_V_SQRT_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = sqrt(S0.f16)
# --- compiled pseudocode ---
D0.f16 = sqrt(S0.f16)
# --- end pseudocode ---
def _VOP1Op_V_RSQ_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = 16'1.0 / sqrt(S0.f16)
# --- compiled pseudocode ---
D0.f16 = 1.0 / sqrt(S0.f16)
# --- end pseudocode ---
def _VOP1Op_V_LOG_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = log2(S0.f16)
# --- compiled pseudocode ---
D0.f16 = log2(S0.f16)
# --- end pseudocode ---
def _VOP1Op_V_EXP_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = pow(16'2.0, S0.f16)
# --- compiled pseudocode ---
D0.f16 = pow(2.0, S0.f16)
# --- end pseudocode ---
def _VOP1Op_V_FREXP_MANT_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if ((64'F(S0.f16) == +INF) || (64'F(S0.f16) == -INF) || isNAN(64'F(S0.f16))) then
# D0.f16 = S0.f16
# else
# D0.f16 = mantissa(S0.f16)
# endif
# --- compiled pseudocode ---
if ((F(S0.f16) == INF) or (F(S0.f16) == (-INF)) or isNAN(F(S0.f16))):
D0.f16 = S0.f16
else:
D0.f16 = mantissa(S0.f16)
# --- end pseudocode ---
def _VOP1Op_V_FREXP_EXP_I16_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if ((64'F(S0.f16) == +INF) || (64'F(S0.f16) == -INF) || isNAN(64'F(S0.f16))) then
# D0.i16 = 16'0
# else
# D0.i16 = 16'I(exponent(S0.f16) - 15 + 1)
# endif
# --- compiled pseudocode ---
if ((F(S0.f16) == INF) or (F(S0.f16) == (-INF)) or isNAN(F(S0.f16))):
D0.i16 = 0
else:
D0.i16 = (exponent(S0.f16) - 15 + 1)
# --- end pseudocode ---
def _VOP1Op_V_FLOOR_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = trunc(S0.f16);
# if ((S0.f16 < 16'0.0) && (S0.f16 != D0.f16)) then
# D0.f16 += -16'1.0
# endif
# --- compiled pseudocode ---
D0.f16 = trunc(S0.f16)
if ((S0.f16 < 0.0) and (S0.f16 != D0.f16)):
D0.f16 += -1.0
# --- end pseudocode ---
def _VOP1Op_V_CEIL_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = trunc(S0.f16);
# if ((S0.f16 > 16'0.0) && (S0.f16 != D0.f16)) then
# D0.f16 += 16'1.0
# endif
# --- compiled pseudocode ---
D0.f16 = trunc(S0.f16)
if ((S0.f16 > 0.0) and (S0.f16 != D0.f16)):
D0.f16 += 1.0
# --- end pseudocode ---
def _VOP1Op_V_TRUNC_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = trunc(S0.f16)
# --- compiled pseudocode ---
D0.f16 = trunc(S0.f16)
# --- end pseudocode ---
def _VOP1Op_V_RNDNE_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = floor(S0.f16 + 16'0.5);
# if (isEven(64'F(floor(S0.f16))) && (fract(S0.f16) == 16'0.5)) then
# D0.f16 -= 16'1.0
# endif
# --- compiled pseudocode ---
D0.f16 = floor(S0.f16 + 0.5)
if (isEven(F(floor(S0.f16))) and (fract(S0.f16) == 0.5)):
D0.f16 -= 1.0
# --- end pseudocode ---
def _VOP1Op_V_FRACT_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = S0.f16 + -floor(S0.f16)
# --- compiled pseudocode ---
D0.f16 = S0.f16 + -floor(S0.f16)
# --- end pseudocode ---
def _VOP1Op_V_SIN_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = sin(S0.f16 * 16'F(PI * 2.0))
# --- compiled pseudocode ---
D0.f16 = sin(S0.f16 * F(PI * 2.0))
# --- end pseudocode ---
def _VOP1Op_V_COS_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = cos(S0.f16 * 16'F(PI * 2.0))
# --- compiled pseudocode ---
D0.f16 = cos(S0.f16 * F(PI * 2.0))
# --- end pseudocode ---
def _VOP1Op_V_CVT_NORM_I16_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i16 = f16_to_snorm(S0.f16)
# --- compiled pseudocode ---
D0.i16 = f16_to_snorm(S0.f16)
# --- end pseudocode ---
def _VOP1Op_V_CVT_NORM_U16_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u16 = f16_to_unorm(S0.f16)
# --- compiled pseudocode ---
D0.u16 = f16_to_unorm(S0.f16)
# --- end pseudocode ---
def _VOP1Op_V_SWAP_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = D0.b32;
# D0.b32 = S0.b32;
# S0.b32 = tmp
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(D0.b32)
D0.b32 = S0.b32
S0.b32 = tmp
# --- end pseudocode ---
def _VOP1Op_V_SWAP_B16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = D0.b16;
# D0.b16 = S0.b16;
# S0.b16 = tmp
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(D0.b16)
D0.b16 = S0.b16
S0.b16 = tmp
# --- end pseudocode ---
def _VOP1Op_V_NOT_B16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u16 = ~S0.u16
# --- compiled pseudocode ---
D0.u16 = ~S0.u16
# --- end pseudocode ---
def _VOP1Op_V_CVT_I32_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = 32'I(signext(S0.i16))
# --- compiled pseudocode ---
D0.i32 = (signext(S0.i16))
# --- end pseudocode ---
def _VOP1Op_V_CVT_U32_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0 = { 16'0, S0.u16 }
# --- compiled pseudocode ---
D0.b32 = _pack(0, S0.u16)
# --- end pseudocode ---
VOP1Op_FUNCTIONS = {
VOP1Op.V_MOV_B32: _VOP1Op_V_MOV_B32,
VOP1Op.V_READFIRSTLANE_B32: _VOP1Op_V_READFIRSTLANE_B32,
VOP1Op.V_CVT_I32_F64: _VOP1Op_V_CVT_I32_F64,
VOP1Op.V_CVT_F64_I32: _VOP1Op_V_CVT_F64_I32,
VOP1Op.V_CVT_F32_I32: _VOP1Op_V_CVT_F32_I32,
VOP1Op.V_CVT_F32_U32: _VOP1Op_V_CVT_F32_U32,
VOP1Op.V_CVT_U32_F32: _VOP1Op_V_CVT_U32_F32,
VOP1Op.V_CVT_I32_F32: _VOP1Op_V_CVT_I32_F32,
VOP1Op.V_CVT_F16_F32: _VOP1Op_V_CVT_F16_F32,
VOP1Op.V_CVT_F32_F16: _VOP1Op_V_CVT_F32_F16,
VOP1Op.V_CVT_NEAREST_I32_F32: _VOP1Op_V_CVT_NEAREST_I32_F32,
VOP1Op.V_CVT_FLOOR_I32_F32: _VOP1Op_V_CVT_FLOOR_I32_F32,
VOP1Op.V_CVT_F32_F64: _VOP1Op_V_CVT_F32_F64,
VOP1Op.V_CVT_F64_F32: _VOP1Op_V_CVT_F64_F32,
VOP1Op.V_CVT_F32_UBYTE0: _VOP1Op_V_CVT_F32_UBYTE0,
VOP1Op.V_CVT_F32_UBYTE1: _VOP1Op_V_CVT_F32_UBYTE1,
VOP1Op.V_CVT_F32_UBYTE2: _VOP1Op_V_CVT_F32_UBYTE2,
VOP1Op.V_CVT_F32_UBYTE3: _VOP1Op_V_CVT_F32_UBYTE3,
VOP1Op.V_CVT_U32_F64: _VOP1Op_V_CVT_U32_F64,
VOP1Op.V_CVT_F64_U32: _VOP1Op_V_CVT_F64_U32,
VOP1Op.V_TRUNC_F64: _VOP1Op_V_TRUNC_F64,
VOP1Op.V_CEIL_F64: _VOP1Op_V_CEIL_F64,
VOP1Op.V_RNDNE_F64: _VOP1Op_V_RNDNE_F64,
VOP1Op.V_FLOOR_F64: _VOP1Op_V_FLOOR_F64,
VOP1Op.V_MOV_B16: _VOP1Op_V_MOV_B16,
VOP1Op.V_FRACT_F32: _VOP1Op_V_FRACT_F32,
VOP1Op.V_TRUNC_F32: _VOP1Op_V_TRUNC_F32,
VOP1Op.V_CEIL_F32: _VOP1Op_V_CEIL_F32,
VOP1Op.V_RNDNE_F32: _VOP1Op_V_RNDNE_F32,
VOP1Op.V_FLOOR_F32: _VOP1Op_V_FLOOR_F32,
VOP1Op.V_EXP_F32: _VOP1Op_V_EXP_F32,
VOP1Op.V_LOG_F32: _VOP1Op_V_LOG_F32,
VOP1Op.V_RCP_F32: _VOP1Op_V_RCP_F32,
VOP1Op.V_RCP_IFLAG_F32: _VOP1Op_V_RCP_IFLAG_F32,
VOP1Op.V_RSQ_F32: _VOP1Op_V_RSQ_F32,
VOP1Op.V_RCP_F64: _VOP1Op_V_RCP_F64,
VOP1Op.V_RSQ_F64: _VOP1Op_V_RSQ_F64,
VOP1Op.V_SQRT_F32: _VOP1Op_V_SQRT_F32,
VOP1Op.V_SQRT_F64: _VOP1Op_V_SQRT_F64,
VOP1Op.V_SIN_F32: _VOP1Op_V_SIN_F32,
VOP1Op.V_COS_F32: _VOP1Op_V_COS_F32,
VOP1Op.V_NOT_B32: _VOP1Op_V_NOT_B32,
VOP1Op.V_BFREV_B32: _VOP1Op_V_BFREV_B32,
VOP1Op.V_CLZ_I32_U32: _VOP1Op_V_CLZ_I32_U32,
VOP1Op.V_CTZ_I32_B32: _VOP1Op_V_CTZ_I32_B32,
VOP1Op.V_CLS_I32: _VOP1Op_V_CLS_I32,
VOP1Op.V_FREXP_EXP_I32_F64: _VOP1Op_V_FREXP_EXP_I32_F64,
VOP1Op.V_FREXP_MANT_F64: _VOP1Op_V_FREXP_MANT_F64,
VOP1Op.V_FRACT_F64: _VOP1Op_V_FRACT_F64,
VOP1Op.V_FREXP_EXP_I32_F32: _VOP1Op_V_FREXP_EXP_I32_F32,
VOP1Op.V_FREXP_MANT_F32: _VOP1Op_V_FREXP_MANT_F32,
VOP1Op.V_MOVRELS_B32: _VOP1Op_V_MOVRELS_B32,
VOP1Op.V_CVT_F16_U16: _VOP1Op_V_CVT_F16_U16,
VOP1Op.V_CVT_F16_I16: _VOP1Op_V_CVT_F16_I16,
VOP1Op.V_CVT_U16_F16: _VOP1Op_V_CVT_U16_F16,
VOP1Op.V_CVT_I16_F16: _VOP1Op_V_CVT_I16_F16,
VOP1Op.V_RCP_F16: _VOP1Op_V_RCP_F16,
VOP1Op.V_SQRT_F16: _VOP1Op_V_SQRT_F16,
VOP1Op.V_RSQ_F16: _VOP1Op_V_RSQ_F16,
VOP1Op.V_LOG_F16: _VOP1Op_V_LOG_F16,
VOP1Op.V_EXP_F16: _VOP1Op_V_EXP_F16,
VOP1Op.V_FREXP_MANT_F16: _VOP1Op_V_FREXP_MANT_F16,
VOP1Op.V_FREXP_EXP_I16_F16: _VOP1Op_V_FREXP_EXP_I16_F16,
VOP1Op.V_FLOOR_F16: _VOP1Op_V_FLOOR_F16,
VOP1Op.V_CEIL_F16: _VOP1Op_V_CEIL_F16,
VOP1Op.V_TRUNC_F16: _VOP1Op_V_TRUNC_F16,
VOP1Op.V_RNDNE_F16: _VOP1Op_V_RNDNE_F16,
VOP1Op.V_FRACT_F16: _VOP1Op_V_FRACT_F16,
VOP1Op.V_SIN_F16: _VOP1Op_V_SIN_F16,
VOP1Op.V_COS_F16: _VOP1Op_V_COS_F16,
VOP1Op.V_CVT_NORM_I16_F16: _VOP1Op_V_CVT_NORM_I16_F16,
VOP1Op.V_CVT_NORM_U16_F16: _VOP1Op_V_CVT_NORM_U16_F16,
VOP1Op.V_SWAP_B32: _VOP1Op_V_SWAP_B32,
VOP1Op.V_SWAP_B16: _VOP1Op_V_SWAP_B16,
VOP1Op.V_NOT_B16: _VOP1Op_V_NOT_B16,
VOP1Op.V_CVT_I32_I16: _VOP1Op_V_CVT_I32_I16,
VOP1Op.V_CVT_U32_U16: _VOP1Op_V_CVT_U32_U16,
}
def _VOP2Op_V_CNDMASK_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = VCC.u64[laneId] ? S1.u32 : S0.u32
# --- compiled pseudocode ---
D0.u32 = ((S1.u32) if (VCC.u64[laneId]) else (S0.u32))
# --- end pseudocode ---
def _VOP2Op_V_DOT2ACC_F32_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = D0.f32;
# tmp += f16_to_f32(S0[15 : 0].f16) * f16_to_f32(S1[15 : 0].f16);
# tmp += f16_to_f32(S0[31 : 16].f16) * f16_to_f32(S1[31 : 16].f16);
# D0.f32 = tmp
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(D0.f32)
tmp += f16_to_f32(S0[15 : 0].f16) * f16_to_f32(S1[15 : 0].f16)
tmp += f16_to_f32(S0[31 : 16].f16) * f16_to_f32(S1[31 : 16].f16)
D0.f32 = tmp
# --- end pseudocode ---
def _VOP2Op_V_ADD_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = S0.f32 + S1.f32
# --- compiled pseudocode ---
D0.f32 = S0.f32 + S1.f32
# --- end pseudocode ---
def _VOP2Op_V_SUB_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = S0.f32 - S1.f32
# --- compiled pseudocode ---
D0.f32 = S0.f32 - S1.f32
# --- end pseudocode ---
def _VOP2Op_V_SUBREV_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = S1.f32 - S0.f32
# --- compiled pseudocode ---
D0.f32 = S1.f32 - S0.f32
# --- end pseudocode ---
def _VOP2Op_V_FMAC_DX9_ZERO_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if ((64'F(S0.f32) == 0.0) || (64'F(S1.f32) == 0.0)) then
# // DX9 rules, 0.0 * x = 0.0
# D0.f32 = S2.f32
# else
# D0.f32 = fma(S0.f32, S1.f32, D0.f32)
# endif
# --- compiled pseudocode ---
if ((F(S0.f32) == 0.0) or (F(S1.f32) == 0.0)):
D0.f32 = S2.f32
else:
D0.f32 = fma(S0.f32, S1.f32, D0.f32)
# --- end pseudocode ---
def _VOP2Op_V_MUL_DX9_ZERO_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if ((64'F(S0.f32) == 0.0) || (64'F(S1.f32) == 0.0)) then
# // DX9 rules, 0.0 * x = 0.0
# D0.f32 = 0.0F
# else
# D0.f32 = S0.f32 * S1.f32
# endif
# --- compiled pseudocode ---
if ((F(S0.f32) == 0.0) or (F(S1.f32) == 0.0)):
D0.f32 = 0.0
else:
D0.f32 = S0.f32 * S1.f32
# --- end pseudocode ---
def _VOP2Op_V_MUL_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = S0.f32 * S1.f32
# --- compiled pseudocode ---
D0.f32 = S0.f32 * S1.f32
# --- end pseudocode ---
def _VOP2Op_V_MUL_I32_I24(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = 32'I(S0.i24) * 32'I(S1.i24)
# --- compiled pseudocode ---
D0.i32 = (S0.i24) * (S1.i24)
# --- end pseudocode ---
def _VOP2Op_V_MUL_HI_I32_I24(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = 32'I((64'I(S0.i24) * 64'I(S1.i24)) >> 32U)
# --- compiled pseudocode ---
D0.i32 = (((S0.i24) * (S1.i24)) >> 32)
# --- end pseudocode ---
def _VOP2Op_V_MUL_U32_U24(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = 32'U(S0.u24) * 32'U(S1.u24)
# --- compiled pseudocode ---
D0.u32 = (S0.u24) * (S1.u24)
# --- end pseudocode ---
def _VOP2Op_V_MUL_HI_U32_U24(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = 32'U((64'U(S0.u24) * 64'U(S1.u24)) >> 32U)
# --- compiled pseudocode ---
D0.u32 = (((S0.u24) * (S1.u24)) >> 32)
# --- end pseudocode ---
def _VOP2Op_V_MIN_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# // Version of comparison where -0.0 < +0.0, differs from IEEE
# if WAVE_MODE.IEEE then
# if isSignalNAN(64'F(S0.f32)) then
# D0.f32 = 32'F(cvtToQuietNAN(64'F(S0.f32)))
# elsif isSignalNAN(64'F(S1.f32)) then
# D0.f32 = 32'F(cvtToQuietNAN(64'F(S1.f32)))
# elsif isQuietNAN(64'F(S1.f32)) then
# D0.f32 = S0.f32
# elsif isQuietNAN(64'F(S0.f32)) then
# D0.f32 = S1.f32
# elsif LT_NEG_ZERO(S0.f32, S1.f32) then
# // NOTE: -0<+0 is TRUE in this comparison
# D0.f32 = S0.f32
# else
# D0.f32 = S1.f32
# endif
# else
# if isNAN(64'F(S1.f32)) then
# D0.f32 = S0.f32
# elsif isNAN(64'F(S0.f32)) then
# D0.f32 = S1.f32
# elsif LT_NEG_ZERO(S0.f32, S1.f32) then
# // NOTE: -0<+0 is TRUE in this comparison
# D0.f32 = S0.f32
# else
# D0.f32 = S1.f32
# endif
# endif;
# // Inequalities in the above pseudocode behave differently from IEEE
# --- compiled pseudocode ---
if WAVE_MODE.IEEE:
if isSignalNAN(F(S0.f32)):
D0.f32 = F(cvtToQuietNAN(F(S0.f32)))
elif isSignalNAN(F(S1.f32)):
D0.f32 = F(cvtToQuietNAN(F(S1.f32)))
elif isQuietNAN(F(S1.f32)):
D0.f32 = S0.f32
elif isQuietNAN(F(S0.f32)):
D0.f32 = S1.f32
elif LT_NEG_ZERO(S0.f32, S1.f32):
D0.f32 = S0.f32
else:
D0.f32 = S1.f32
else:
if isNAN(F(S1.f32)):
D0.f32 = S0.f32
elif isNAN(F(S0.f32)):
D0.f32 = S1.f32
elif LT_NEG_ZERO(S0.f32, S1.f32):
D0.f32 = S0.f32
else:
D0.f32 = S1.f32
# --- end pseudocode ---
def _VOP2Op_V_MAX_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# // Version of comparison where +0.0 > -0.0, differs from IEEE
# if WAVE_MODE.IEEE then
# if isSignalNAN(64'F(S0.f32)) then
# D0.f32 = 32'F(cvtToQuietNAN(64'F(S0.f32)))
# elsif isSignalNAN(64'F(S1.f32)) then
# D0.f32 = 32'F(cvtToQuietNAN(64'F(S1.f32)))
# elsif isQuietNAN(64'F(S1.f32)) then
# D0.f32 = S0.f32
# elsif isQuietNAN(64'F(S0.f32)) then
# D0.f32 = S1.f32
# elsif GT_NEG_ZERO(S0.f32, S1.f32) then
# // NOTE: +0>-0 is TRUE in this comparison
# D0.f32 = S0.f32
# else
# D0.f32 = S1.f32
# endif
# else
# if isNAN(64'F(S1.f32)) then
# D0.f32 = S0.f32
# elsif isNAN(64'F(S0.f32)) then
# D0.f32 = S1.f32
# elsif GT_NEG_ZERO(S0.f32, S1.f32) then
# // NOTE: +0>-0 is TRUE in this comparison
# D0.f32 = S0.f32
# else
# D0.f32 = S1.f32
# endif
# endif;
# // Inequalities in the above pseudocode behave differently from IEEE
# --- compiled pseudocode ---
if WAVE_MODE.IEEE:
if isSignalNAN(F(S0.f32)):
D0.f32 = F(cvtToQuietNAN(F(S0.f32)))
elif isSignalNAN(F(S1.f32)):
D0.f32 = F(cvtToQuietNAN(F(S1.f32)))
elif isQuietNAN(F(S1.f32)):
D0.f32 = S0.f32
elif isQuietNAN(F(S0.f32)):
D0.f32 = S1.f32
elif GT_NEG_ZERO(S0.f32, S1.f32):
D0.f32 = S0.f32
else:
D0.f32 = S1.f32
else:
if isNAN(F(S1.f32)):
D0.f32 = S0.f32
elif isNAN(F(S0.f32)):
D0.f32 = S1.f32
elif GT_NEG_ZERO(S0.f32, S1.f32):
D0.f32 = S0.f32
else:
D0.f32 = S1.f32
# --- end pseudocode ---
def _VOP2Op_V_MIN_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = S0.i32 < S1.i32 ? S0.i32 : S1.i32
# --- compiled pseudocode ---
D0.i32 = ((S0.i32) if (S0.i32 < S1.i32) else (S1.i32))
# --- end pseudocode ---
def _VOP2Op_V_MAX_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = S0.i32 >= S1.i32 ? S0.i32 : S1.i32
# --- compiled pseudocode ---
D0.i32 = ((S0.i32) if (S0.i32 >= S1.i32) else (S1.i32))
# --- end pseudocode ---
def _VOP2Op_V_MIN_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = S0.u32 < S1.u32 ? S0.u32 : S1.u32
# --- compiled pseudocode ---
D0.u32 = ((S0.u32) if (S0.u32 < S1.u32) else (S1.u32))
# --- end pseudocode ---
def _VOP2Op_V_MAX_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = S0.u32 >= S1.u32 ? S0.u32 : S1.u32
# --- compiled pseudocode ---
D0.u32 = ((S0.u32) if (S0.u32 >= S1.u32) else (S1.u32))
# --- end pseudocode ---
def _VOP2Op_V_LSHLREV_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = (S1.u32 << S0[4 : 0].u32)
# --- compiled pseudocode ---
D0.u32 = (S1.u32 << S0[4 : 0].u32)
# --- end pseudocode ---
def _VOP2Op_V_LSHRREV_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = (S1.u32 >> S0[4 : 0].u32)
# --- compiled pseudocode ---
D0.u32 = (S1.u32 >> S0[4 : 0].u32)
# --- end pseudocode ---
def _VOP2Op_V_ASHRREV_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = (S1.i32 >> S0[4 : 0].u32)
# --- compiled pseudocode ---
D0.i32 = (S1.i32 >> S0[4 : 0].u32)
# --- end pseudocode ---
def _VOP2Op_V_AND_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = (S0.u32 & S1.u32)
# --- compiled pseudocode ---
D0.u32 = (S0.u32 & S1.u32)
# --- end pseudocode ---
def _VOP2Op_V_OR_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = (S0.u32 | S1.u32)
# --- compiled pseudocode ---
D0.u32 = (S0.u32 | S1.u32)
# --- end pseudocode ---
def _VOP2Op_V_XOR_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = (S0.u32 ^ S1.u32)
# --- compiled pseudocode ---
D0.u32 = (S0.u32 ^ S1.u32)
# --- end pseudocode ---
def _VOP2Op_V_XNOR_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = ~(S0.u32 ^ S1.u32)
# --- compiled pseudocode ---
D0.u32 = ~(S0.u32 ^ S1.u32)
# --- end pseudocode ---
def _VOP2Op_V_ADD_CO_CI_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = 64'U(S0.u32) + 64'U(S1.u32) + VCC.u64[laneId].u64;
# VCC.u64[laneId] = tmp >= 0x100000000ULL ? 1'1U : 1'0U;
# // VCC is an UNSIGNED overflow/carry-out for V_ADD_CO_CI_U32.
# D0.u32 = tmp.u32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg((S0.u32) + (S1.u32) + VCC.u64[laneId])
VCC.u64[laneId] = ((1) if (tmp >= 0x100000000) else (0))
D0.u32 = tmp.u32
# --- end pseudocode ---
def _VOP2Op_V_SUB_CO_CI_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = S0.u32 - S1.u32 - VCC.u64[laneId].u32;
# VCC.u64[laneId] = 64'U(S1.u32) + VCC.u64[laneId].u64 > 64'U(S0.u32) ? 1'1U : 1'0U;
# // VCC is an UNSIGNED overflow/carry-out for V_SUB_CO_CI_U32.
# D0.u32 = tmp.u32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(S0.u32 - S1.u32 - VCC.u64[laneId])
VCC.u64[laneId] = ((1) if ((S1.u32) + VCC.u64[laneId] > (S0.u32)) else (0))
D0.u32 = tmp.u32
# --- end pseudocode ---
def _VOP2Op_V_SUBREV_CO_CI_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = S1.u32 - S0.u32 - VCC.u64[laneId].u32;
# VCC.u64[laneId] = 64'U(S0.u32) + VCC.u64[laneId].u64 > 64'U(S1.u32) ? 1'1U : 1'0U;
# // VCC is an UNSIGNED overflow/carry-out for V_SUB_CO_CI_U32.
# D0.u32 = tmp.u32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(S1.u32 - S0.u32 - VCC.u64[laneId])
VCC.u64[laneId] = ((1) if ((S0.u32) + VCC.u64[laneId] > (S1.u32)) else (0))
D0.u32 = tmp.u32
# --- end pseudocode ---
def _VOP2Op_V_ADD_NC_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = S0.u32 + S1.u32
# --- compiled pseudocode ---
D0.u32 = S0.u32 + S1.u32
# --- end pseudocode ---
def _VOP2Op_V_SUB_NC_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = S0.u32 - S1.u32
# --- compiled pseudocode ---
D0.u32 = S0.u32 - S1.u32
# --- end pseudocode ---
def _VOP2Op_V_SUBREV_NC_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = S1.u32 - S0.u32
# --- compiled pseudocode ---
D0.u32 = S1.u32 - S0.u32
# --- end pseudocode ---
def _VOP2Op_V_FMAC_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = fma(S0.f32, S1.f32, D0.f32)
# --- compiled pseudocode ---
D0.f32 = fma(S0.f32, S1.f32, D0.f32)
# --- end pseudocode ---
def _VOP2Op_V_FMAMK_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = fma(S0.f32, SIMM32.f32, S1.f32)
SIMM32 = SIMM16
# --- compiled pseudocode ---
D0.f32 = fma(S0.f32, SIMM32.f32, S1.f32)
# --- end pseudocode ---
def _VOP2Op_V_FMAAK_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = fma(S0.f32, S1.f32, SIMM32.f32)
SIMM32 = SIMM16
# --- compiled pseudocode ---
D0.f32 = fma(S0.f32, S1.f32, SIMM32.f32)
# --- end pseudocode ---
def _VOP2Op_V_CVT_PK_RTZ_F16_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# prev_mode = ROUND_MODE;
# tmp[15 : 0].f16 = f32_to_f16(S0.f32);
# tmp[31 : 16].f16 = f32_to_f16(S1.f32);
tmp = Reg(0)
# --- compiled pseudocode ---
prev_mode = ROUND_MODE
tmp[15 : 0].f16 = f32_to_f16(S0.f32)
tmp[31 : 16].f16 = f32_to_f16(S1.f32)
# --- end pseudocode ---
def _VOP2Op_V_ADD_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = S0.f16 + S1.f16
# --- compiled pseudocode ---
D0.f16 = S0.f16 + S1.f16
# --- end pseudocode ---
def _VOP2Op_V_SUB_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = S0.f16 - S1.f16
# --- compiled pseudocode ---
D0.f16 = S0.f16 - S1.f16
# --- end pseudocode ---
def _VOP2Op_V_SUBREV_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = S1.f16 - S0.f16
# --- compiled pseudocode ---
D0.f16 = S1.f16 - S0.f16
# --- end pseudocode ---
def _VOP2Op_V_MUL_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = S0.f16 * S1.f16
# --- compiled pseudocode ---
D0.f16 = S0.f16 * S1.f16
# --- end pseudocode ---
def _VOP2Op_V_FMAC_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = fma(S0.f16, S1.f16, D0.f16)
# --- compiled pseudocode ---
D0.f16 = fma(S0.f16, S1.f16, D0.f16)
# --- end pseudocode ---
def _VOP2Op_V_FMAMK_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = fma(S0.f16, SIMM32.f16, S1.f16)
SIMM32 = SIMM16
# --- compiled pseudocode ---
D0.f16 = fma(S0.f16, SIMM32.f16, S1.f16)
# --- end pseudocode ---
def _VOP2Op_V_FMAAK_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = fma(S0.f16, S1.f16, SIMM32.f16)
SIMM32 = SIMM16
# --- compiled pseudocode ---
D0.f16 = fma(S0.f16, S1.f16, SIMM32.f16)
# --- end pseudocode ---
def _VOP2Op_V_MAX_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# // Version of comparison where +0.0 > -0.0, differs from IEEE
# if WAVE_MODE.IEEE then
# if isSignalNAN(64'F(S0.f16)) then
# D0.f16 = 16'F(cvtToQuietNAN(64'F(S0.f16)))
# elsif isSignalNAN(64'F(S1.f16)) then
# D0.f16 = 16'F(cvtToQuietNAN(64'F(S1.f16)))
# elsif isQuietNAN(64'F(S1.f16)) then
# D0.f16 = S0.f16
# elsif isQuietNAN(64'F(S0.f16)) then
# D0.f16 = S1.f16
# elsif GT_NEG_ZERO(S0.f16, S1.f16) then
# // NOTE: +0>-0 is TRUE in this comparison
# D0.f16 = S0.f16
# else
# D0.f16 = S1.f16
# endif
# else
# if isNAN(64'F(S1.f16)) then
# D0.f16 = S0.f16
# elsif isNAN(64'F(S0.f16)) then
# D0.f16 = S1.f16
# elsif GT_NEG_ZERO(S0.f16, S1.f16) then
# // NOTE: +0>-0 is TRUE in this comparison
# D0.f16 = S0.f16
# else
# D0.f16 = S1.f16
# endif
# endif;
# // Inequalities in the above pseudocode behave differently from IEEE
# --- compiled pseudocode ---
if WAVE_MODE.IEEE:
if isSignalNAN(F(S0.f16)):
D0.f16 = F(cvtToQuietNAN(F(S0.f16)))
elif isSignalNAN(F(S1.f16)):
D0.f16 = F(cvtToQuietNAN(F(S1.f16)))
elif isQuietNAN(F(S1.f16)):
D0.f16 = S0.f16
elif isQuietNAN(F(S0.f16)):
D0.f16 = S1.f16
elif GT_NEG_ZERO(S0.f16, S1.f16):
D0.f16 = S0.f16
else:
D0.f16 = S1.f16
else:
if isNAN(F(S1.f16)):
D0.f16 = S0.f16
elif isNAN(F(S0.f16)):
D0.f16 = S1.f16
elif GT_NEG_ZERO(S0.f16, S1.f16):
D0.f16 = S0.f16
else:
D0.f16 = S1.f16
# --- end pseudocode ---
def _VOP2Op_V_MIN_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# // Version of comparison where -0.0 < +0.0, differs from IEEE
# if WAVE_MODE.IEEE then
# if isSignalNAN(64'F(S0.f16)) then
# D0.f16 = 16'F(cvtToQuietNAN(64'F(S0.f16)))
# elsif isSignalNAN(64'F(S1.f16)) then
# D0.f16 = 16'F(cvtToQuietNAN(64'F(S1.f16)))
# elsif isQuietNAN(64'F(S1.f16)) then
# D0.f16 = S0.f16
# elsif isQuietNAN(64'F(S0.f16)) then
# D0.f16 = S1.f16
# elsif LT_NEG_ZERO(S0.f16, S1.f16) then
# // NOTE: -0<+0 is TRUE in this comparison
# D0.f16 = S0.f16
# else
# D0.f16 = S1.f16
# endif
# else
# if isNAN(64'F(S1.f16)) then
# D0.f16 = S0.f16
# elsif isNAN(64'F(S0.f16)) then
# D0.f16 = S1.f16
# elsif LT_NEG_ZERO(S0.f16, S1.f16) then
# // NOTE: -0<+0 is TRUE in this comparison
# D0.f16 = S0.f16
# else
# D0.f16 = S1.f16
# endif
# endif;
# // Inequalities in the above pseudocode behave differently from IEEE
# --- compiled pseudocode ---
if WAVE_MODE.IEEE:
if isSignalNAN(F(S0.f16)):
D0.f16 = F(cvtToQuietNAN(F(S0.f16)))
elif isSignalNAN(F(S1.f16)):
D0.f16 = F(cvtToQuietNAN(F(S1.f16)))
elif isQuietNAN(F(S1.f16)):
D0.f16 = S0.f16
elif isQuietNAN(F(S0.f16)):
D0.f16 = S1.f16
elif LT_NEG_ZERO(S0.f16, S1.f16):
D0.f16 = S0.f16
else:
D0.f16 = S1.f16
else:
if isNAN(F(S1.f16)):
D0.f16 = S0.f16
elif isNAN(F(S0.f16)):
D0.f16 = S1.f16
elif LT_NEG_ZERO(S0.f16, S1.f16):
D0.f16 = S0.f16
else:
D0.f16 = S1.f16
# --- end pseudocode ---
def _VOP2Op_V_LDEXP_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = S0.f16 * 16'F(2.0F ** 32'I(S1.i16))
# --- compiled pseudocode ---
D0.f16 = S0.f16 * F(2.0 ** (S1.i16))
# --- end pseudocode ---
def _VOP2Op_V_PK_FMAC_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0[31 : 16].f16 = fma(S0[31 : 16].f16, S1[31 : 16].f16, D0[31 : 16].f16);
# D0[15 : 0].f16 = fma(S0[15 : 0].f16, S1[15 : 0].f16, D0[15 : 0].f16)
# --- compiled pseudocode ---
D0[31 : 16].f16 = fma(S0[31 : 16].f16, S1[31 : 16].f16, D0[31 : 16].f16)
D0[15 : 0].f16 = fma(S0[15 : 0].f16, S1[15 : 0].f16, D0[15 : 0].f16)
# --- end pseudocode ---
VOP2Op_FUNCTIONS = {
VOP2Op.V_CNDMASK_B32: _VOP2Op_V_CNDMASK_B32,
VOP2Op.V_DOT2ACC_F32_F16: _VOP2Op_V_DOT2ACC_F32_F16,
VOP2Op.V_ADD_F32: _VOP2Op_V_ADD_F32,
VOP2Op.V_SUB_F32: _VOP2Op_V_SUB_F32,
VOP2Op.V_SUBREV_F32: _VOP2Op_V_SUBREV_F32,
VOP2Op.V_FMAC_DX9_ZERO_F32: _VOP2Op_V_FMAC_DX9_ZERO_F32,
VOP2Op.V_MUL_DX9_ZERO_F32: _VOP2Op_V_MUL_DX9_ZERO_F32,
VOP2Op.V_MUL_F32: _VOP2Op_V_MUL_F32,
VOP2Op.V_MUL_I32_I24: _VOP2Op_V_MUL_I32_I24,
VOP2Op.V_MUL_HI_I32_I24: _VOP2Op_V_MUL_HI_I32_I24,
VOP2Op.V_MUL_U32_U24: _VOP2Op_V_MUL_U32_U24,
VOP2Op.V_MUL_HI_U32_U24: _VOP2Op_V_MUL_HI_U32_U24,
VOP2Op.V_MIN_F32: _VOP2Op_V_MIN_F32,
VOP2Op.V_MAX_F32: _VOP2Op_V_MAX_F32,
VOP2Op.V_MIN_I32: _VOP2Op_V_MIN_I32,
VOP2Op.V_MAX_I32: _VOP2Op_V_MAX_I32,
VOP2Op.V_MIN_U32: _VOP2Op_V_MIN_U32,
VOP2Op.V_MAX_U32: _VOP2Op_V_MAX_U32,
VOP2Op.V_LSHLREV_B32: _VOP2Op_V_LSHLREV_B32,
VOP2Op.V_LSHRREV_B32: _VOP2Op_V_LSHRREV_B32,
VOP2Op.V_ASHRREV_I32: _VOP2Op_V_ASHRREV_I32,
VOP2Op.V_AND_B32: _VOP2Op_V_AND_B32,
VOP2Op.V_OR_B32: _VOP2Op_V_OR_B32,
VOP2Op.V_XOR_B32: _VOP2Op_V_XOR_B32,
VOP2Op.V_XNOR_B32: _VOP2Op_V_XNOR_B32,
VOP2Op.V_ADD_CO_CI_U32: _VOP2Op_V_ADD_CO_CI_U32,
VOP2Op.V_SUB_CO_CI_U32: _VOP2Op_V_SUB_CO_CI_U32,
VOP2Op.V_SUBREV_CO_CI_U32: _VOP2Op_V_SUBREV_CO_CI_U32,
VOP2Op.V_ADD_NC_U32: _VOP2Op_V_ADD_NC_U32,
VOP2Op.V_SUB_NC_U32: _VOP2Op_V_SUB_NC_U32,
VOP2Op.V_SUBREV_NC_U32: _VOP2Op_V_SUBREV_NC_U32,
VOP2Op.V_FMAC_F32: _VOP2Op_V_FMAC_F32,
VOP2Op.V_FMAMK_F32: _VOP2Op_V_FMAMK_F32,
VOP2Op.V_FMAAK_F32: _VOP2Op_V_FMAAK_F32,
VOP2Op.V_CVT_PK_RTZ_F16_F32: _VOP2Op_V_CVT_PK_RTZ_F16_F32,
VOP2Op.V_ADD_F16: _VOP2Op_V_ADD_F16,
VOP2Op.V_SUB_F16: _VOP2Op_V_SUB_F16,
VOP2Op.V_SUBREV_F16: _VOP2Op_V_SUBREV_F16,
VOP2Op.V_MUL_F16: _VOP2Op_V_MUL_F16,
VOP2Op.V_FMAC_F16: _VOP2Op_V_FMAC_F16,
VOP2Op.V_FMAMK_F16: _VOP2Op_V_FMAMK_F16,
VOP2Op.V_FMAAK_F16: _VOP2Op_V_FMAAK_F16,
VOP2Op.V_MAX_F16: _VOP2Op_V_MAX_F16,
VOP2Op.V_MIN_F16: _VOP2Op_V_MIN_F16,
VOP2Op.V_LDEXP_F16: _VOP2Op_V_LDEXP_F16,
VOP2Op.V_PK_FMAC_F16: _VOP2Op_V_PK_FMAC_F16,
}
def _VOP3Op_V_CMP_F_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 0. Store the result into VCC or a scalar register.
# D0.u64[laneId] = 1'0U;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = 0
# --- end pseudocode ---
def _VOP3Op_V_CMP_LT_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is less than the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.f16 < S1.f16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f16 < S1.f16
# --- end pseudocode ---
def _VOP3Op_V_CMP_EQ_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.f16 == S1.f16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f16 == S1.f16
# --- end pseudocode ---
def _VOP3Op_V_CMP_LE_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.f16 <= S1.f16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f16 <= S1.f16
# --- end pseudocode ---
def _VOP3Op_V_CMP_GT_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is greater than the second input. Store the result into VCC
# D0.u64[laneId] = S0.f16 > S1.f16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f16 > S1.f16
# --- end pseudocode ---
def _VOP3Op_V_CMP_LG_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.f16 <> S1.f16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f16 != S1.f16
# --- end pseudocode ---
def _VOP3Op_V_CMP_GE_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.f16 >= S1.f16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f16 >= S1.f16
# --- end pseudocode ---
def _VOP3Op_V_CMP_O_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is orderable to the second input. Store the result into VCC
# D0.u64[laneId] = (!isNAN(64'F(S0.f16)) && !isNAN(64'F(S1.f16)));
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = ( not isNAN(F(S0.f16)) and not isNAN(F(S1.f16)))
# --- end pseudocode ---
def _VOP3Op_V_CMP_U_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# VCC or a scalar register.
# D0.u64[laneId] = (isNAN(64'F(S0.f16)) || isNAN(64'F(S1.f16)));
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = (isNAN(F(S0.f16)) or isNAN(F(S1.f16)))
# --- end pseudocode ---
def _VOP3Op_V_CMP_NGE_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = !(S0.f16 >= S1.f16);
# // With NAN inputs this is not the same operation as <
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f16 >= S1.f16)
# --- end pseudocode ---
def _VOP3Op_V_CMP_NLG_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = !(S0.f16 <> S1.f16);
# // With NAN inputs this is not the same operation as ==
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f16 != S1.f16)
# --- end pseudocode ---
def _VOP3Op_V_CMP_NGT_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# VCC or a scalar register.
# D0.u64[laneId] = !(S0.f16 > S1.f16);
# // With NAN inputs this is not the same operation as <=
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f16 > S1.f16)
# --- end pseudocode ---
def _VOP3Op_V_CMP_NLE_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = !(S0.f16 <= S1.f16);
# // With NAN inputs this is not the same operation as >
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f16 <= S1.f16)
# --- end pseudocode ---
def _VOP3Op_V_CMP_NEQ_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is not equal to the second input. Store the result into VCC
# D0.u64[laneId] = !(S0.f16 == S1.f16);
# // With NAN inputs this is not the same operation as !=
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f16 == S1.f16)
# --- end pseudocode ---
def _VOP3Op_V_CMP_NLT_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is not less than the second input. Store the result into VCC
# D0.u64[laneId] = !(S0.f16 < S1.f16);
# // With NAN inputs this is not the same operation as >=
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f16 < S1.f16)
# --- end pseudocode ---
def _VOP3Op_V_CMP_T_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1. Store the result into VCC or a scalar register.
# D0.u64[laneId] = 1'1U;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = 1
# --- end pseudocode ---
def _VOP3Op_V_CMP_F_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 0. Store the result into VCC or a scalar register.
# D0.u64[laneId] = 1'0U;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = 0
# --- end pseudocode ---
def _VOP3Op_V_CMP_LT_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is less than the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.f32 < S1.f32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f32 < S1.f32
# --- end pseudocode ---
def _VOP3Op_V_CMP_EQ_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.f32 == S1.f32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f32 == S1.f32
# --- end pseudocode ---
def _VOP3Op_V_CMP_LE_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.f32 <= S1.f32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f32 <= S1.f32
# --- end pseudocode ---
def _VOP3Op_V_CMP_GT_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is greater than the second input. Store the result into VCC
# D0.u64[laneId] = S0.f32 > S1.f32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f32 > S1.f32
# --- end pseudocode ---
def _VOP3Op_V_CMP_LG_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.f32 <> S1.f32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f32 != S1.f32
# --- end pseudocode ---
def _VOP3Op_V_CMP_GE_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.f32 >= S1.f32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f32 >= S1.f32
# --- end pseudocode ---
def _VOP3Op_V_CMP_O_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is orderable to the second input. Store the result into VCC
# D0.u64[laneId] = (!isNAN(64'F(S0.f32)) && !isNAN(64'F(S1.f32)));
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = ( not isNAN(F(S0.f32)) and not isNAN(F(S1.f32)))
# --- end pseudocode ---
def _VOP3Op_V_CMP_U_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# VCC or a scalar register.
# D0.u64[laneId] = (isNAN(64'F(S0.f32)) || isNAN(64'F(S1.f32)));
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = (isNAN(F(S0.f32)) or isNAN(F(S1.f32)))
# --- end pseudocode ---
def _VOP3Op_V_CMP_NGE_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = !(S0.f32 >= S1.f32);
# // With NAN inputs this is not the same operation as <
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f32 >= S1.f32)
# --- end pseudocode ---
def _VOP3Op_V_CMP_NLG_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = !(S0.f32 <> S1.f32);
# // With NAN inputs this is not the same operation as ==
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f32 != S1.f32)
# --- end pseudocode ---
def _VOP3Op_V_CMP_NGT_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# VCC or a scalar register.
# D0.u64[laneId] = !(S0.f32 > S1.f32);
# // With NAN inputs this is not the same operation as <=
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f32 > S1.f32)
# --- end pseudocode ---
def _VOP3Op_V_CMP_NLE_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = !(S0.f32 <= S1.f32);
# // With NAN inputs this is not the same operation as >
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f32 <= S1.f32)
# --- end pseudocode ---
def _VOP3Op_V_CMP_NEQ_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is not equal to the second input. Store the result into VCC
# D0.u64[laneId] = !(S0.f32 == S1.f32);
# // With NAN inputs this is not the same operation as !=
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f32 == S1.f32)
# --- end pseudocode ---
def _VOP3Op_V_CMP_NLT_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is not less than the second input. Store the result into VCC
# D0.u64[laneId] = !(S0.f32 < S1.f32);
# // With NAN inputs this is not the same operation as >=
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f32 < S1.f32)
# --- end pseudocode ---
def _VOP3Op_V_CMP_T_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1. Store the result into VCC or a scalar register.
# D0.u64[laneId] = 1'1U;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = 1
# --- end pseudocode ---
def _VOP3Op_V_CMP_F_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 0. Store the result into VCC or a scalar register.
# D0.u64[laneId] = 1'0U;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = 0
# --- end pseudocode ---
def _VOP3Op_V_CMP_LT_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is less than the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.f64 < S1.f64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f64 < S1.f64
# --- end pseudocode ---
def _VOP3Op_V_CMP_EQ_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.f64 == S1.f64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f64 == S1.f64
# --- end pseudocode ---
def _VOP3Op_V_CMP_LE_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.f64 <= S1.f64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f64 <= S1.f64
# --- end pseudocode ---
def _VOP3Op_V_CMP_GT_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is greater than the second input. Store the result into VCC
# D0.u64[laneId] = S0.f64 > S1.f64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f64 > S1.f64
# --- end pseudocode ---
def _VOP3Op_V_CMP_LG_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.f64 <> S1.f64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f64 != S1.f64
# --- end pseudocode ---
def _VOP3Op_V_CMP_GE_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.f64 >= S1.f64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f64 >= S1.f64
# --- end pseudocode ---
def _VOP3Op_V_CMP_O_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is orderable to the second input. Store the result into VCC
# D0.u64[laneId] = (!isNAN(S0.f64) && !isNAN(S1.f64));
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = ( not isNAN(S0.f64) and not isNAN(S1.f64))
# --- end pseudocode ---
def _VOP3Op_V_CMP_U_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# VCC or a scalar register.
# D0.u64[laneId] = (isNAN(S0.f64) || isNAN(S1.f64));
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = (isNAN(S0.f64) or isNAN(S1.f64))
# --- end pseudocode ---
def _VOP3Op_V_CMP_NGE_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = !(S0.f64 >= S1.f64);
# // With NAN inputs this is not the same operation as <
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f64 >= S1.f64)
# --- end pseudocode ---
def _VOP3Op_V_CMP_NLG_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = !(S0.f64 <> S1.f64);
# // With NAN inputs this is not the same operation as ==
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f64 != S1.f64)
# --- end pseudocode ---
def _VOP3Op_V_CMP_NGT_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# VCC or a scalar register.
# D0.u64[laneId] = !(S0.f64 > S1.f64);
# // With NAN inputs this is not the same operation as <=
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f64 > S1.f64)
# --- end pseudocode ---
def _VOP3Op_V_CMP_NLE_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = !(S0.f64 <= S1.f64);
# // With NAN inputs this is not the same operation as >
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f64 <= S1.f64)
# --- end pseudocode ---
def _VOP3Op_V_CMP_NEQ_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is not equal to the second input. Store the result into VCC
# D0.u64[laneId] = !(S0.f64 == S1.f64);
# // With NAN inputs this is not the same operation as !=
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f64 == S1.f64)
# --- end pseudocode ---
def _VOP3Op_V_CMP_NLT_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is not less than the second input. Store the result into VCC
# D0.u64[laneId] = !(S0.f64 < S1.f64);
# // With NAN inputs this is not the same operation as >=
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f64 < S1.f64)
# --- end pseudocode ---
def _VOP3Op_V_CMP_T_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1. Store the result into VCC or a scalar register.
# D0.u64[laneId] = 1'1U;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = 1
# --- end pseudocode ---
def _VOP3Op_V_CMP_LT_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is less than the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.i16 < S1.i16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i16 < S1.i16
# --- end pseudocode ---
def _VOP3Op_V_CMP_EQ_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.i16 == S1.i16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i16 == S1.i16
# --- end pseudocode ---
def _VOP3Op_V_CMP_LE_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.i16 <= S1.i16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i16 <= S1.i16
# --- end pseudocode ---
def _VOP3Op_V_CMP_GT_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is greater than the second input. Store the result into VCC
# D0.u64[laneId] = S0.i16 > S1.i16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i16 > S1.i16
# --- end pseudocode ---
def _VOP3Op_V_CMP_NE_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is not equal to the second input. Store the result into VCC
# D0.u64[laneId] = S0.i16 <> S1.i16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i16 != S1.i16
# --- end pseudocode ---
def _VOP3Op_V_CMP_GE_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.i16 >= S1.i16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i16 >= S1.i16
# --- end pseudocode ---
def _VOP3Op_V_CMP_LT_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is less than the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.u16 < S1.u16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u16 < S1.u16
# --- end pseudocode ---
def _VOP3Op_V_CMP_EQ_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.u16 == S1.u16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u16 == S1.u16
# --- end pseudocode ---
def _VOP3Op_V_CMP_LE_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.u16 <= S1.u16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u16 <= S1.u16
# --- end pseudocode ---
def _VOP3Op_V_CMP_GT_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is greater than the second input. Store the result into VCC
# D0.u64[laneId] = S0.u16 > S1.u16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u16 > S1.u16
# --- end pseudocode ---
def _VOP3Op_V_CMP_NE_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is not equal to the second input. Store the result into VCC
# D0.u64[laneId] = S0.u16 <> S1.u16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u16 != S1.u16
# --- end pseudocode ---
def _VOP3Op_V_CMP_GE_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.u16 >= S1.u16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u16 >= S1.u16
# --- end pseudocode ---
def _VOP3Op_V_CMP_F_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 0. Store the result into VCC or a scalar register.
# D0.u64[laneId] = 1'0U;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = 0
# --- end pseudocode ---
def _VOP3Op_V_CMP_LT_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is less than the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.i32 < S1.i32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i32 < S1.i32
# --- end pseudocode ---
def _VOP3Op_V_CMP_EQ_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.i32 == S1.i32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i32 == S1.i32
# --- end pseudocode ---
def _VOP3Op_V_CMP_LE_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.i32 <= S1.i32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i32 <= S1.i32
# --- end pseudocode ---
def _VOP3Op_V_CMP_GT_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is greater than the second input. Store the result into VCC
# D0.u64[laneId] = S0.i32 > S1.i32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i32 > S1.i32
# --- end pseudocode ---
def _VOP3Op_V_CMP_NE_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is not equal to the second input. Store the result into VCC
# D0.u64[laneId] = S0.i32 <> S1.i32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i32 != S1.i32
# --- end pseudocode ---
def _VOP3Op_V_CMP_GE_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.i32 >= S1.i32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i32 >= S1.i32
# --- end pseudocode ---
def _VOP3Op_V_CMP_T_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1. Store the result into VCC or a scalar register.
# D0.u64[laneId] = 1'1U;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = 1
# --- end pseudocode ---
def _VOP3Op_V_CMP_F_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 0. Store the result into VCC or a scalar register.
# D0.u64[laneId] = 1'0U;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = 0
# --- end pseudocode ---
def _VOP3Op_V_CMP_LT_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is less than the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.u32 < S1.u32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u32 < S1.u32
# --- end pseudocode ---
def _VOP3Op_V_CMP_EQ_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.u32 == S1.u32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u32 == S1.u32
# --- end pseudocode ---
def _VOP3Op_V_CMP_LE_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.u32 <= S1.u32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u32 <= S1.u32
# --- end pseudocode ---
def _VOP3Op_V_CMP_GT_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is greater than the second input. Store the result into VCC
# D0.u64[laneId] = S0.u32 > S1.u32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u32 > S1.u32
# --- end pseudocode ---
def _VOP3Op_V_CMP_NE_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is not equal to the second input. Store the result into VCC
# D0.u64[laneId] = S0.u32 <> S1.u32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u32 != S1.u32
# --- end pseudocode ---
def _VOP3Op_V_CMP_GE_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.u32 >= S1.u32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u32 >= S1.u32
# --- end pseudocode ---
def _VOP3Op_V_CMP_T_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1. Store the result into VCC or a scalar register.
# D0.u64[laneId] = 1'1U;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = 1
# --- end pseudocode ---
def _VOP3Op_V_CMP_F_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 0. Store the result into VCC or a scalar register.
# D0.u64[laneId] = 1'0U;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = 0
# --- end pseudocode ---
def _VOP3Op_V_CMP_LT_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is less than the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.i64 < S1.i64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i64 < S1.i64
# --- end pseudocode ---
def _VOP3Op_V_CMP_EQ_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.i64 == S1.i64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i64 == S1.i64
# --- end pseudocode ---
def _VOP3Op_V_CMP_LE_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.i64 <= S1.i64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i64 <= S1.i64
# --- end pseudocode ---
def _VOP3Op_V_CMP_GT_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is greater than the second input. Store the result into VCC
# D0.u64[laneId] = S0.i64 > S1.i64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i64 > S1.i64
# --- end pseudocode ---
def _VOP3Op_V_CMP_NE_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is not equal to the second input. Store the result into VCC
# D0.u64[laneId] = S0.i64 <> S1.i64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i64 != S1.i64
# --- end pseudocode ---
def _VOP3Op_V_CMP_GE_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.i64 >= S1.i64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i64 >= S1.i64
# --- end pseudocode ---
def _VOP3Op_V_CMP_T_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1. Store the result into VCC or a scalar register.
# D0.u64[laneId] = 1'1U;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = 1
# --- end pseudocode ---
def _VOP3Op_V_CMP_F_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 0. Store the result into VCC or a scalar register.
# D0.u64[laneId] = 1'0U;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = 0
# --- end pseudocode ---
def _VOP3Op_V_CMP_LT_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is less than the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.u64 < S1.u64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u64 < S1.u64
# --- end pseudocode ---
def _VOP3Op_V_CMP_EQ_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.u64 == S1.u64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u64 == S1.u64
# --- end pseudocode ---
def _VOP3Op_V_CMP_LE_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.u64 <= S1.u64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u64 <= S1.u64
# --- end pseudocode ---
def _VOP3Op_V_CMP_GT_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is greater than the second input. Store the result into VCC
# D0.u64[laneId] = S0.u64 > S1.u64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u64 > S1.u64
# --- end pseudocode ---
def _VOP3Op_V_CMP_NE_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is not equal to the second input. Store the result into VCC
# D0.u64[laneId] = S0.u64 <> S1.u64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u64 != S1.u64
# --- end pseudocode ---
def _VOP3Op_V_CMP_GE_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.u64 >= S1.u64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u64 >= S1.u64
# --- end pseudocode ---
def _VOP3Op_V_CMP_T_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1. Store the result into VCC or a scalar register.
# D0.u64[laneId] = 1'1U;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = 1
# --- end pseudocode ---
def _VOP3Op_V_CMP_CLASS_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# half-precision float, and set the per-lane condition code to the result. Store the result into VCC or a scalar
# S1.u[0] value is a signaling NAN.
# S1.u[1] value is a quiet NAN.
# S1.u[2] value is negative infinity.
# S1.u[3] value is a negative normal value.
# S1.u[4] value is a negative denormal value.
# S1.u[5] value is negative zero.
# S1.u[6] value is positive zero.
# S1.u[7] value is a positive denormal value.
# S1.u[8] value is a positive normal value.
# S1.u[9] value is positive infinity.
# declare result : 1'U;
# if isSignalNAN(64'F(S0.f16)) then
# result = S1.u32[0]
# elsif isQuietNAN(64'F(S0.f16)) then
# result = S1.u32[1]
# elsif exponent(S0.f16) == 31 then
# // +-INF
# result = S1.u32[sign(S0.f16) ? 2 : 9]
# elsif exponent(S0.f16) > 0 then
# // +-normal value
# result = S1.u32[sign(S0.f16) ? 3 : 8]
# elsif 64'F(abs(S0.f16)) > 0.0 then
# // +-denormal value
# result = S1.u32[sign(S0.f16) ? 4 : 7]
# else
# // +-0.0
# result = S1.u32[sign(S0.f16) ? 5 : 6]
# endif;
# D0.u64[laneId] = result;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
if isSignalNAN(F(S0.f16)):
result = S1.u32[0]
elif isQuietNAN(F(S0.f16)):
result = S1.u32[1]
elif exponent(S0.f16) == 31:
result = S1.u32[((2) if (sign(S0.f16)) else (9))]
elif exponent(S0.f16) > 0:
result = S1.u32[((3) if (sign(S0.f16)) else (8))]
elif F(abs(S0.f16)) > 0.0:
result = S1.u32[((4) if (sign(S0.f16)) else (7))]
else:
result = S1.u32[((5) if (sign(S0.f16)) else (6))]
D0.u64[laneId] = result
# --- end pseudocode ---
def _VOP3Op_V_CMP_CLASS_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# single-precision float, and set the per-lane condition code to the result. Store the result into VCC or a scalar
# S1.u[0] value is a signaling NAN.
# S1.u[1] value is a quiet NAN.
# S1.u[2] value is negative infinity.
# S1.u[3] value is a negative normal value.
# S1.u[4] value is a negative denormal value.
# S1.u[5] value is negative zero.
# S1.u[6] value is positive zero.
# S1.u[7] value is a positive denormal value.
# S1.u[8] value is a positive normal value.
# S1.u[9] value is positive infinity.
# declare result : 1'U;
# if isSignalNAN(64'F(S0.f32)) then
# result = S1.u32[0]
# elsif isQuietNAN(64'F(S0.f32)) then
# result = S1.u32[1]
# elsif exponent(S0.f32) == 255 then
# // +-INF
# result = S1.u32[sign(S0.f32) ? 2 : 9]
# elsif exponent(S0.f32) > 0 then
# // +-normal value
# result = S1.u32[sign(S0.f32) ? 3 : 8]
# elsif 64'F(abs(S0.f32)) > 0.0 then
# // +-denormal value
# result = S1.u32[sign(S0.f32) ? 4 : 7]
# else
# // +-0.0
# result = S1.u32[sign(S0.f32) ? 5 : 6]
# endif;
# D0.u64[laneId] = result;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
if isSignalNAN(F(S0.f32)):
result = S1.u32[0]
elif isQuietNAN(F(S0.f32)):
result = S1.u32[1]
elif exponent(S0.f32) == 255:
result = S1.u32[((2) if (sign(S0.f32)) else (9))]
elif exponent(S0.f32) > 0:
result = S1.u32[((3) if (sign(S0.f32)) else (8))]
elif F(abs(S0.f32)) > 0.0:
result = S1.u32[((4) if (sign(S0.f32)) else (7))]
else:
result = S1.u32[((5) if (sign(S0.f32)) else (6))]
D0.u64[laneId] = result
# --- end pseudocode ---
def _VOP3Op_V_CMP_CLASS_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# double-precision float, and set the per-lane condition code to the result. Store the result into VCC or a scalar
# S1.u[0] value is a signaling NAN.
# S1.u[1] value is a quiet NAN.
# S1.u[2] value is negative infinity.
# S1.u[3] value is a negative normal value.
# S1.u[4] value is a negative denormal value.
# S1.u[5] value is negative zero.
# S1.u[6] value is positive zero.
# S1.u[7] value is a positive denormal value.
# S1.u[8] value is a positive normal value.
# S1.u[9] value is positive infinity.
# declare result : 1'U;
# if isSignalNAN(S0.f64) then
# result = S1.u32[0]
# elsif isQuietNAN(S0.f64) then
# result = S1.u32[1]
# elsif exponent(S0.f64) == 2047 then
# // +-INF
# result = S1.u32[sign(S0.f64) ? 2 : 9]
# elsif exponent(S0.f64) > 0 then
# // +-normal value
# result = S1.u32[sign(S0.f64) ? 3 : 8]
# elsif abs(S0.f64) > 0.0 then
# // +-denormal value
# result = S1.u32[sign(S0.f64) ? 4 : 7]
# else
# // +-0.0
# result = S1.u32[sign(S0.f64) ? 5 : 6]
# endif;
# D0.u64[laneId] = result;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
if isSignalNAN(S0.f64):
result = S1.u32[0]
elif isQuietNAN(S0.f64):
result = S1.u32[1]
elif exponent(S0.f64) == 2047:
result = S1.u32[((2) if (sign(S0.f64)) else (9))]
elif exponent(S0.f64) > 0:
result = S1.u32[((3) if (sign(S0.f64)) else (8))]
elif abs(S0.f64) > 0.0:
result = S1.u32[((4) if (sign(S0.f64)) else (7))]
else:
result = S1.u32[((5) if (sign(S0.f64)) else (6))]
D0.u64[laneId] = result
# --- end pseudocode ---
def _VOP3Op_V_CMPX_F_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = 1'0U
# --- compiled pseudocode ---
EXEC.u64[laneId] = 0
# --- end pseudocode ---
def _VOP3Op_V_CMPX_LT_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f16 < S1.f16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f16 < S1.f16
# --- end pseudocode ---
def _VOP3Op_V_CMPX_EQ_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into the EXEC
# EXEC.u64[laneId] = S0.f16 == S1.f16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f16 == S1.f16
# --- end pseudocode ---
def _VOP3Op_V_CMPX_LE_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f16 <= S1.f16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f16 <= S1.f16
# --- end pseudocode ---
def _VOP3Op_V_CMPX_GT_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f16 > S1.f16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f16 > S1.f16
# --- end pseudocode ---
def _VOP3Op_V_CMPX_LG_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f16 <> S1.f16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f16 != S1.f16
# --- end pseudocode ---
def _VOP3Op_V_CMPX_GE_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f16 >= S1.f16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f16 >= S1.f16
# --- end pseudocode ---
def _VOP3Op_V_CMPX_O_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = (!isNAN(64'F(S0.f16)) && !isNAN(64'F(S1.f16)))
# --- compiled pseudocode ---
EXEC.u64[laneId] = ( not isNAN(F(S0.f16)) and not isNAN(F(S1.f16)))
# --- end pseudocode ---
def _VOP3Op_V_CMPX_U_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = (isNAN(64'F(S0.f16)) || isNAN(64'F(S1.f16)))
# --- compiled pseudocode ---
EXEC.u64[laneId] = (isNAN(F(S0.f16)) or isNAN(F(S1.f16)))
# --- end pseudocode ---
def _VOP3Op_V_CMPX_NGE_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f16 >= S1.f16);
# // With NAN inputs this is not the same operation as <
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f16 >= S1.f16)
# --- end pseudocode ---
def _VOP3Op_V_CMPX_NLG_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f16 <> S1.f16);
# // With NAN inputs this is not the same operation as ==
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f16 != S1.f16)
# --- end pseudocode ---
def _VOP3Op_V_CMPX_NGT_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f16 > S1.f16);
# // With NAN inputs this is not the same operation as <=
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f16 > S1.f16)
# --- end pseudocode ---
def _VOP3Op_V_CMPX_NLE_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f16 <= S1.f16);
# // With NAN inputs this is not the same operation as >
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f16 <= S1.f16)
# --- end pseudocode ---
def _VOP3Op_V_CMPX_NEQ_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f16 == S1.f16);
# // With NAN inputs this is not the same operation as !=
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f16 == S1.f16)
# --- end pseudocode ---
def _VOP3Op_V_CMPX_NLT_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f16 < S1.f16);
# // With NAN inputs this is not the same operation as >=
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f16 < S1.f16)
# --- end pseudocode ---
def _VOP3Op_V_CMPX_T_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = 1'1U
# --- compiled pseudocode ---
EXEC.u64[laneId] = 1
# --- end pseudocode ---
def _VOP3Op_V_CMPX_F_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = 1'0U
# --- compiled pseudocode ---
EXEC.u64[laneId] = 0
# --- end pseudocode ---
def _VOP3Op_V_CMPX_LT_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f32 < S1.f32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f32 < S1.f32
# --- end pseudocode ---
def _VOP3Op_V_CMPX_EQ_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into the EXEC
# EXEC.u64[laneId] = S0.f32 == S1.f32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f32 == S1.f32
# --- end pseudocode ---
def _VOP3Op_V_CMPX_LE_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f32 <= S1.f32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f32 <= S1.f32
# --- end pseudocode ---
def _VOP3Op_V_CMPX_GT_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f32 > S1.f32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f32 > S1.f32
# --- end pseudocode ---
def _VOP3Op_V_CMPX_LG_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f32 <> S1.f32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f32 != S1.f32
# --- end pseudocode ---
def _VOP3Op_V_CMPX_GE_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f32 >= S1.f32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f32 >= S1.f32
# --- end pseudocode ---
def _VOP3Op_V_CMPX_O_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = (!isNAN(64'F(S0.f32)) && !isNAN(64'F(S1.f32)))
# --- compiled pseudocode ---
EXEC.u64[laneId] = ( not isNAN(F(S0.f32)) and not isNAN(F(S1.f32)))
# --- end pseudocode ---
def _VOP3Op_V_CMPX_U_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = (isNAN(64'F(S0.f32)) || isNAN(64'F(S1.f32)))
# --- compiled pseudocode ---
EXEC.u64[laneId] = (isNAN(F(S0.f32)) or isNAN(F(S1.f32)))
# --- end pseudocode ---
def _VOP3Op_V_CMPX_NGE_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f32 >= S1.f32);
# // With NAN inputs this is not the same operation as <
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f32 >= S1.f32)
# --- end pseudocode ---
def _VOP3Op_V_CMPX_NLG_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f32 <> S1.f32);
# // With NAN inputs this is not the same operation as ==
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f32 != S1.f32)
# --- end pseudocode ---
def _VOP3Op_V_CMPX_NGT_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f32 > S1.f32);
# // With NAN inputs this is not the same operation as <=
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f32 > S1.f32)
# --- end pseudocode ---
def _VOP3Op_V_CMPX_NLE_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f32 <= S1.f32);
# // With NAN inputs this is not the same operation as >
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f32 <= S1.f32)
# --- end pseudocode ---
def _VOP3Op_V_CMPX_NEQ_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f32 == S1.f32);
# // With NAN inputs this is not the same operation as !=
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f32 == S1.f32)
# --- end pseudocode ---
def _VOP3Op_V_CMPX_NLT_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f32 < S1.f32);
# // With NAN inputs this is not the same operation as >=
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f32 < S1.f32)
# --- end pseudocode ---
def _VOP3Op_V_CMPX_T_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = 1'1U
# --- compiled pseudocode ---
EXEC.u64[laneId] = 1
# --- end pseudocode ---
def _VOP3Op_V_CMPX_F_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = 1'0U
# --- compiled pseudocode ---
EXEC.u64[laneId] = 0
# --- end pseudocode ---
def _VOP3Op_V_CMPX_LT_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f64 < S1.f64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f64 < S1.f64
# --- end pseudocode ---
def _VOP3Op_V_CMPX_EQ_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into the EXEC
# EXEC.u64[laneId] = S0.f64 == S1.f64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f64 == S1.f64
# --- end pseudocode ---
def _VOP3Op_V_CMPX_LE_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f64 <= S1.f64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f64 <= S1.f64
# --- end pseudocode ---
def _VOP3Op_V_CMPX_GT_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f64 > S1.f64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f64 > S1.f64
# --- end pseudocode ---
def _VOP3Op_V_CMPX_LG_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f64 <> S1.f64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f64 != S1.f64
# --- end pseudocode ---
def _VOP3Op_V_CMPX_GE_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f64 >= S1.f64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f64 >= S1.f64
# --- end pseudocode ---
def _VOP3Op_V_CMPX_O_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = (!isNAN(S0.f64) && !isNAN(S1.f64))
# --- compiled pseudocode ---
EXEC.u64[laneId] = ( not isNAN(S0.f64) and not isNAN(S1.f64))
# --- end pseudocode ---
def _VOP3Op_V_CMPX_U_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = (isNAN(S0.f64) || isNAN(S1.f64))
# --- compiled pseudocode ---
EXEC.u64[laneId] = (isNAN(S0.f64) or isNAN(S1.f64))
# --- end pseudocode ---
def _VOP3Op_V_CMPX_NGE_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f64 >= S1.f64);
# // With NAN inputs this is not the same operation as <
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f64 >= S1.f64)
# --- end pseudocode ---
def _VOP3Op_V_CMPX_NLG_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f64 <> S1.f64);
# // With NAN inputs this is not the same operation as ==
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f64 != S1.f64)
# --- end pseudocode ---
def _VOP3Op_V_CMPX_NGT_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f64 > S1.f64);
# // With NAN inputs this is not the same operation as <=
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f64 > S1.f64)
# --- end pseudocode ---
def _VOP3Op_V_CMPX_NLE_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f64 <= S1.f64);
# // With NAN inputs this is not the same operation as >
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f64 <= S1.f64)
# --- end pseudocode ---
def _VOP3Op_V_CMPX_NEQ_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f64 == S1.f64);
# // With NAN inputs this is not the same operation as !=
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f64 == S1.f64)
# --- end pseudocode ---
def _VOP3Op_V_CMPX_NLT_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f64 < S1.f64);
# // With NAN inputs this is not the same operation as >=
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f64 < S1.f64)
# --- end pseudocode ---
def _VOP3Op_V_CMPX_T_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = 1'1U
# --- compiled pseudocode ---
EXEC.u64[laneId] = 1
# --- end pseudocode ---
def _VOP3Op_V_CMPX_LT_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i16 < S1.i16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i16 < S1.i16
# --- end pseudocode ---
def _VOP3Op_V_CMPX_EQ_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into the EXEC
# EXEC.u64[laneId] = S0.i16 == S1.i16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i16 == S1.i16
# --- end pseudocode ---
def _VOP3Op_V_CMPX_LE_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i16 <= S1.i16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i16 <= S1.i16
# --- end pseudocode ---
def _VOP3Op_V_CMPX_GT_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i16 > S1.i16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i16 > S1.i16
# --- end pseudocode ---
def _VOP3Op_V_CMPX_NE_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i16 <> S1.i16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i16 != S1.i16
# --- end pseudocode ---
def _VOP3Op_V_CMPX_GE_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i16 >= S1.i16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i16 >= S1.i16
# --- end pseudocode ---
def _VOP3Op_V_CMPX_LT_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u16 < S1.u16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u16 < S1.u16
# --- end pseudocode ---
def _VOP3Op_V_CMPX_EQ_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into the EXEC
# EXEC.u64[laneId] = S0.u16 == S1.u16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u16 == S1.u16
# --- end pseudocode ---
def _VOP3Op_V_CMPX_LE_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u16 <= S1.u16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u16 <= S1.u16
# --- end pseudocode ---
def _VOP3Op_V_CMPX_GT_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u16 > S1.u16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u16 > S1.u16
# --- end pseudocode ---
def _VOP3Op_V_CMPX_NE_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u16 <> S1.u16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u16 != S1.u16
# --- end pseudocode ---
def _VOP3Op_V_CMPX_GE_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u16 >= S1.u16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u16 >= S1.u16
# --- end pseudocode ---
def _VOP3Op_V_CMPX_F_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = 1'0U
# --- compiled pseudocode ---
EXEC.u64[laneId] = 0
# --- end pseudocode ---
def _VOP3Op_V_CMPX_LT_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i32 < S1.i32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i32 < S1.i32
# --- end pseudocode ---
def _VOP3Op_V_CMPX_EQ_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into the EXEC
# EXEC.u64[laneId] = S0.i32 == S1.i32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i32 == S1.i32
# --- end pseudocode ---
def _VOP3Op_V_CMPX_LE_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i32 <= S1.i32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i32 <= S1.i32
# --- end pseudocode ---
def _VOP3Op_V_CMPX_GT_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i32 > S1.i32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i32 > S1.i32
# --- end pseudocode ---
def _VOP3Op_V_CMPX_NE_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i32 <> S1.i32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i32 != S1.i32
# --- end pseudocode ---
def _VOP3Op_V_CMPX_GE_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i32 >= S1.i32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i32 >= S1.i32
# --- end pseudocode ---
def _VOP3Op_V_CMPX_T_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = 1'1U
# --- compiled pseudocode ---
EXEC.u64[laneId] = 1
# --- end pseudocode ---
def _VOP3Op_V_CMPX_F_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = 1'0U
# --- compiled pseudocode ---
EXEC.u64[laneId] = 0
# --- end pseudocode ---
def _VOP3Op_V_CMPX_LT_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u32 < S1.u32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u32 < S1.u32
# --- end pseudocode ---
def _VOP3Op_V_CMPX_EQ_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into the EXEC
# EXEC.u64[laneId] = S0.u32 == S1.u32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u32 == S1.u32
# --- end pseudocode ---
def _VOP3Op_V_CMPX_LE_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u32 <= S1.u32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u32 <= S1.u32
# --- end pseudocode ---
def _VOP3Op_V_CMPX_GT_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u32 > S1.u32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u32 > S1.u32
# --- end pseudocode ---
def _VOP3Op_V_CMPX_NE_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u32 <> S1.u32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u32 != S1.u32
# --- end pseudocode ---
def _VOP3Op_V_CMPX_GE_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u32 >= S1.u32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u32 >= S1.u32
# --- end pseudocode ---
def _VOP3Op_V_CMPX_T_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = 1'1U
# --- compiled pseudocode ---
EXEC.u64[laneId] = 1
# --- end pseudocode ---
def _VOP3Op_V_CMPX_F_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = 1'0U
# --- compiled pseudocode ---
EXEC.u64[laneId] = 0
# --- end pseudocode ---
def _VOP3Op_V_CMPX_LT_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i64 < S1.i64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i64 < S1.i64
# --- end pseudocode ---
def _VOP3Op_V_CMPX_EQ_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into the EXEC
# EXEC.u64[laneId] = S0.i64 == S1.i64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i64 == S1.i64
# --- end pseudocode ---
def _VOP3Op_V_CMPX_LE_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i64 <= S1.i64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i64 <= S1.i64
# --- end pseudocode ---
def _VOP3Op_V_CMPX_GT_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i64 > S1.i64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i64 > S1.i64
# --- end pseudocode ---
def _VOP3Op_V_CMPX_NE_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i64 <> S1.i64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i64 != S1.i64
# --- end pseudocode ---
def _VOP3Op_V_CMPX_GE_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i64 >= S1.i64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i64 >= S1.i64
# --- end pseudocode ---
def _VOP3Op_V_CMPX_T_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = 1'1U
# --- compiled pseudocode ---
EXEC.u64[laneId] = 1
# --- end pseudocode ---
def _VOP3Op_V_CMPX_F_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = 1'0U
# --- compiled pseudocode ---
EXEC.u64[laneId] = 0
# --- end pseudocode ---
def _VOP3Op_V_CMPX_LT_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u64 < S1.u64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u64 < S1.u64
# --- end pseudocode ---
def _VOP3Op_V_CMPX_EQ_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into the EXEC
# EXEC.u64[laneId] = S0.u64 == S1.u64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u64 == S1.u64
# --- end pseudocode ---
def _VOP3Op_V_CMPX_LE_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u64 <= S1.u64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u64 <= S1.u64
# --- end pseudocode ---
def _VOP3Op_V_CMPX_GT_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u64 > S1.u64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u64 > S1.u64
# --- end pseudocode ---
def _VOP3Op_V_CMPX_NE_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u64 <> S1.u64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u64 != S1.u64
# --- end pseudocode ---
def _VOP3Op_V_CMPX_GE_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u64 >= S1.u64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u64 >= S1.u64
# --- end pseudocode ---
def _VOP3Op_V_CMPX_T_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = 1'1U
# --- compiled pseudocode ---
EXEC.u64[laneId] = 1
# --- end pseudocode ---
def _VOP3Op_V_CMPX_CLASS_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# S1.u[0] value is a signaling NAN.
# S1.u[1] value is a quiet NAN.
# S1.u[2] value is negative infinity.
# S1.u[3] value is a negative normal value.
# S1.u[4] value is a negative denormal value.
# S1.u[5] value is negative zero.
# S1.u[6] value is positive zero.
# S1.u[7] value is a positive denormal value.
# S1.u[8] value is a positive normal value.
# S1.u[9] value is positive infinity.
# declare result : 1'U;
# if isSignalNAN(64'F(S0.f16)) then
# result = S1.u32[0]
# elsif isQuietNAN(64'F(S0.f16)) then
# result = S1.u32[1]
# elsif exponent(S0.f16) == 31 then
# // +-INF
# result = S1.u32[sign(S0.f16) ? 2 : 9]
# elsif exponent(S0.f16) > 0 then
# // +-normal value
# result = S1.u32[sign(S0.f16) ? 3 : 8]
# elsif 64'F(abs(S0.f16)) > 0.0 then
# // +-denormal value
# result = S1.u32[sign(S0.f16) ? 4 : 7]
# else
# // +-0.0
# result = S1.u32[sign(S0.f16) ? 5 : 6]
# endif;
# EXEC.u64[laneId] = result
# --- compiled pseudocode ---
if isSignalNAN(F(S0.f16)):
result = S1.u32[0]
elif isQuietNAN(F(S0.f16)):
result = S1.u32[1]
elif exponent(S0.f16) == 31:
result = S1.u32[((2) if (sign(S0.f16)) else (9))]
elif exponent(S0.f16) > 0:
result = S1.u32[((3) if (sign(S0.f16)) else (8))]
elif F(abs(S0.f16)) > 0.0:
result = S1.u32[((4) if (sign(S0.f16)) else (7))]
else:
result = S1.u32[((5) if (sign(S0.f16)) else (6))]
EXEC.u64[laneId] = result
# --- end pseudocode ---
def _VOP3Op_V_CMPX_CLASS_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# S1.u[0] value is a signaling NAN.
# S1.u[1] value is a quiet NAN.
# S1.u[2] value is negative infinity.
# S1.u[3] value is a negative normal value.
# S1.u[4] value is a negative denormal value.
# S1.u[5] value is negative zero.
# S1.u[6] value is positive zero.
# S1.u[7] value is a positive denormal value.
# S1.u[8] value is a positive normal value.
# S1.u[9] value is positive infinity.
# declare result : 1'U;
# if isSignalNAN(64'F(S0.f32)) then
# result = S1.u32[0]
# elsif isQuietNAN(64'F(S0.f32)) then
# result = S1.u32[1]
# elsif exponent(S0.f32) == 255 then
# // +-INF
# result = S1.u32[sign(S0.f32) ? 2 : 9]
# elsif exponent(S0.f32) > 0 then
# // +-normal value
# result = S1.u32[sign(S0.f32) ? 3 : 8]
# elsif 64'F(abs(S0.f32)) > 0.0 then
# // +-denormal value
# result = S1.u32[sign(S0.f32) ? 4 : 7]
# else
# // +-0.0
# result = S1.u32[sign(S0.f32) ? 5 : 6]
# endif;
# EXEC.u64[laneId] = result
# --- compiled pseudocode ---
if isSignalNAN(F(S0.f32)):
result = S1.u32[0]
elif isQuietNAN(F(S0.f32)):
result = S1.u32[1]
elif exponent(S0.f32) == 255:
result = S1.u32[((2) if (sign(S0.f32)) else (9))]
elif exponent(S0.f32) > 0:
result = S1.u32[((3) if (sign(S0.f32)) else (8))]
elif F(abs(S0.f32)) > 0.0:
result = S1.u32[((4) if (sign(S0.f32)) else (7))]
else:
result = S1.u32[((5) if (sign(S0.f32)) else (6))]
EXEC.u64[laneId] = result
# --- end pseudocode ---
def _VOP3Op_V_CMPX_CLASS_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# S1.u[0] value is a signaling NAN.
# S1.u[1] value is a quiet NAN.
# S1.u[2] value is negative infinity.
# S1.u[3] value is a negative normal value.
# S1.u[4] value is a negative denormal value.
# S1.u[5] value is negative zero.
# S1.u[6] value is positive zero.
# S1.u[7] value is a positive denormal value.
# S1.u[8] value is a positive normal value.
# S1.u[9] value is positive infinity.
# declare result : 1'U;
# if isSignalNAN(S0.f64) then
# result = S1.u32[0]
# elsif isQuietNAN(S0.f64) then
# result = S1.u32[1]
# elsif exponent(S0.f64) == 2047 then
# // +-INF
# result = S1.u32[sign(S0.f64) ? 2 : 9]
# elsif exponent(S0.f64) > 0 then
# // +-normal value
# result = S1.u32[sign(S0.f64) ? 3 : 8]
# elsif abs(S0.f64) > 0.0 then
# // +-denormal value
# result = S1.u32[sign(S0.f64) ? 4 : 7]
# else
# // +-0.0
# result = S1.u32[sign(S0.f64) ? 5 : 6]
# endif;
# EXEC.u64[laneId] = result
# --- compiled pseudocode ---
if isSignalNAN(S0.f64):
result = S1.u32[0]
elif isQuietNAN(S0.f64):
result = S1.u32[1]
elif exponent(S0.f64) == 2047:
result = S1.u32[((2) if (sign(S0.f64)) else (9))]
elif exponent(S0.f64) > 0:
result = S1.u32[((3) if (sign(S0.f64)) else (8))]
elif abs(S0.f64) > 0.0:
result = S1.u32[((4) if (sign(S0.f64)) else (7))]
else:
result = S1.u32[((5) if (sign(S0.f64)) else (6))]
EXEC.u64[laneId] = result
# --- end pseudocode ---
def _VOP3Op_V_MOV_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.b32 = S0.b32
# --- compiled pseudocode ---
D0.b32 = S0.b32
# --- end pseudocode ---
def _VOP3Op_V_READFIRSTLANE_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# declare lane : 32'U;
# if WAVE64 then
# // 64 lanes
# if EXEC == 0x0LL then
# lane = 0U;
# // Force lane 0 if all lanes are disabled
# else
# lane = 32'U(s_ff1_i32_b64(EXEC));
# // Lowest active lane
# endif
# else
# // 32 lanes
# if EXEC_LO.i32 == 0 then
# lane = 0U;
# // Force lane 0 if all lanes are disabled
# else
# lane = 32'U(s_ff1_i32_b32(EXEC_LO));
# // Lowest active lane
# endif
# endif;
# D0.b32 = VGPR[lane][SRC0.u32]
EXEC_LO = SliceProxy(EXEC, 31, 0)
# --- compiled pseudocode ---
if WAVE64:
if EXEC == 0x0:
lane = 0
else:
lane = (s_ff1_i32_b64(EXEC))
else:
if EXEC_LO.i32 == 0:
lane = 0
else:
lane = (s_ff1_i32_b32(EXEC_LO))
D0.b32 = VGPR[lane][SRC0.u32]
# --- end pseudocode ---
def _VOP3Op_V_CVT_I32_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = f64_to_i32(S0.f64)
# --- compiled pseudocode ---
D0.i32 = f64_to_i32(S0.f64)
# --- end pseudocode ---
def _VOP3Op_V_CVT_F64_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f64 = i32_to_f64(S0.i32)
# --- compiled pseudocode ---
D0.f64 = i32_to_f64(S0.i32)
# --- end pseudocode ---
def _VOP3Op_V_CVT_F32_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = i32_to_f32(S0.i32)
# --- compiled pseudocode ---
D0.f32 = i32_to_f32(S0.i32)
# --- end pseudocode ---
def _VOP3Op_V_CVT_F32_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = u32_to_f32(S0.u32)
# --- compiled pseudocode ---
D0.f32 = u32_to_f32(S0.u32)
# --- end pseudocode ---
def _VOP3Op_V_CVT_U32_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = f32_to_u32(S0.f32)
# --- compiled pseudocode ---
D0.u32 = f32_to_u32(S0.f32)
# --- end pseudocode ---
def _VOP3Op_V_CVT_I32_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = f32_to_i32(S0.f32)
# --- compiled pseudocode ---
D0.i32 = f32_to_i32(S0.f32)
# --- end pseudocode ---
def _VOP3Op_V_CVT_F16_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = f32_to_f16(S0.f32)
# --- compiled pseudocode ---
D0.f16 = f32_to_f16(S0.f32)
# --- end pseudocode ---
def _VOP3Op_V_CVT_F32_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = f16_to_f32(S0.f16)
# --- compiled pseudocode ---
D0.f32 = f16_to_f32(S0.f16)
# --- end pseudocode ---
def _VOP3Op_V_CVT_NEAREST_I32_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = f32_to_i32(floor(S0.f32 + 0.5F))
# --- compiled pseudocode ---
D0.i32 = f32_to_i32(floor(S0.f32 + 0.5))
# --- end pseudocode ---
def _VOP3Op_V_CVT_FLOOR_I32_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = f32_to_i32(floor(S0.f32))
# --- compiled pseudocode ---
D0.i32 = f32_to_i32(floor(S0.f32))
# --- end pseudocode ---
def _VOP3Op_V_CVT_F32_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = f64_to_f32(S0.f64)
# --- compiled pseudocode ---
D0.f32 = f64_to_f32(S0.f64)
# --- end pseudocode ---
def _VOP3Op_V_CVT_F64_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f64 = f32_to_f64(S0.f32)
# --- compiled pseudocode ---
D0.f64 = f32_to_f64(S0.f32)
# --- end pseudocode ---
def _VOP3Op_V_CVT_F32_UBYTE0(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = u32_to_f32(S0[7 : 0].u32)
# --- compiled pseudocode ---
D0.f32 = u32_to_f32(S0[7 : 0].u32)
# --- end pseudocode ---
def _VOP3Op_V_CVT_F32_UBYTE1(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = u32_to_f32(S0[15 : 8].u32)
# --- compiled pseudocode ---
D0.f32 = u32_to_f32(S0[15 : 8].u32)
# --- end pseudocode ---
def _VOP3Op_V_CVT_F32_UBYTE2(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = u32_to_f32(S0[23 : 16].u32)
# --- compiled pseudocode ---
D0.f32 = u32_to_f32(S0[23 : 16].u32)
# --- end pseudocode ---
def _VOP3Op_V_CVT_F32_UBYTE3(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = u32_to_f32(S0[31 : 24].u32)
# --- compiled pseudocode ---
D0.f32 = u32_to_f32(S0[31 : 24].u32)
# --- end pseudocode ---
def _VOP3Op_V_CVT_U32_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = f64_to_u32(S0.f64)
# --- compiled pseudocode ---
D0.u32 = f64_to_u32(S0.f64)
# --- end pseudocode ---
def _VOP3Op_V_CVT_F64_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f64 = u32_to_f64(S0.u32)
# --- compiled pseudocode ---
D0.f64 = u32_to_f64(S0.u32)
# --- end pseudocode ---
def _VOP3Op_V_TRUNC_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f64 = trunc(S0.f64)
# --- compiled pseudocode ---
D0.f64 = trunc(S0.f64)
# --- end pseudocode ---
def _VOP3Op_V_CEIL_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f64 = trunc(S0.f64);
# if ((S0.f64 > 0.0) && (S0.f64 != D0.f64)) then
# D0.f64 += 1.0
# endif
# --- compiled pseudocode ---
D0.f64 = trunc(S0.f64)
if ((S0.f64 > 0.0) and (S0.f64 != D0.f64)):
D0.f64 += 1.0
# --- end pseudocode ---
def _VOP3Op_V_RNDNE_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f64 = floor(S0.f64 + 0.5);
# if (isEven(floor(S0.f64)) && (fract(S0.f64) == 0.5)) then
# D0.f64 -= 1.0
# endif
# --- compiled pseudocode ---
D0.f64 = floor(S0.f64 + 0.5)
if (isEven(floor(S0.f64)) and (fract(S0.f64) == 0.5)):
D0.f64 -= 1.0
# --- end pseudocode ---
def _VOP3Op_V_FLOOR_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f64 = trunc(S0.f64);
# if ((S0.f64 < 0.0) && (S0.f64 != D0.f64)) then
# D0.f64 += -1.0
# endif
# --- compiled pseudocode ---
D0.f64 = trunc(S0.f64)
if ((S0.f64 < 0.0) and (S0.f64 != D0.f64)):
D0.f64 += -1.0
# --- end pseudocode ---
def _VOP3Op_V_MOV_B16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.b16 = S0.b16
# --- compiled pseudocode ---
D0.b16 = S0.b16
# --- end pseudocode ---
def _VOP3Op_V_FRACT_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = S0.f32 + -floor(S0.f32)
# --- compiled pseudocode ---
D0.f32 = S0.f32 + -floor(S0.f32)
# --- end pseudocode ---
def _VOP3Op_V_TRUNC_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = trunc(S0.f32)
# --- compiled pseudocode ---
D0.f32 = trunc(S0.f32)
# --- end pseudocode ---
def _VOP3Op_V_CEIL_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = trunc(S0.f32);
# if ((S0.f32 > 0.0F) && (S0.f32 != D0.f32)) then
# D0.f32 += 1.0F
# endif
# --- compiled pseudocode ---
D0.f32 = trunc(S0.f32)
if ((S0.f32 > 0.0) and (S0.f32 != D0.f32)):
D0.f32 += 1.0
# --- end pseudocode ---
def _VOP3Op_V_RNDNE_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = floor(S0.f32 + 0.5F);
# if (isEven(64'F(floor(S0.f32))) && (fract(S0.f32) == 0.5F)) then
# D0.f32 -= 1.0F
# endif
# --- compiled pseudocode ---
D0.f32 = floor(S0.f32 + 0.5)
if (isEven(F(floor(S0.f32))) and (fract(S0.f32) == 0.5)):
D0.f32 -= 1.0
# --- end pseudocode ---
def _VOP3Op_V_FLOOR_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = trunc(S0.f32);
# if ((S0.f32 < 0.0F) && (S0.f32 != D0.f32)) then
# D0.f32 += -1.0F
# endif
# --- compiled pseudocode ---
D0.f32 = trunc(S0.f32)
if ((S0.f32 < 0.0) and (S0.f32 != D0.f32)):
D0.f32 += -1.0
# --- end pseudocode ---
def _VOP3Op_V_EXP_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = pow(2.0F, S0.f32)
# --- compiled pseudocode ---
D0.f32 = pow(2.0, S0.f32)
# --- end pseudocode ---
def _VOP3Op_V_LOG_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = log2(S0.f32)
# --- compiled pseudocode ---
D0.f32 = log2(S0.f32)
# --- end pseudocode ---
def _VOP3Op_V_RCP_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = 1.0F / S0.f32
# --- compiled pseudocode ---
D0.f32 = 1.0 / S0.f32
# --- end pseudocode ---
def _VOP3Op_V_RCP_IFLAG_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = 1.0F / S0.f32;
# // Can only raise integer DIV_BY_ZERO exception
# --- compiled pseudocode ---
D0.f32 = 1.0 / S0.f32
# --- end pseudocode ---
def _VOP3Op_V_RSQ_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = 1.0F / sqrt(S0.f32)
# --- compiled pseudocode ---
D0.f32 = 1.0 / sqrt(S0.f32)
# --- end pseudocode ---
def _VOP3Op_V_RCP_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f64 = 1.0 / S0.f64
# --- compiled pseudocode ---
D0.f64 = 1.0 / S0.f64
# --- end pseudocode ---
def _VOP3Op_V_RSQ_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f64 = 1.0 / sqrt(S0.f64)
# --- compiled pseudocode ---
D0.f64 = 1.0 / sqrt(S0.f64)
# --- end pseudocode ---
def _VOP3Op_V_SQRT_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = sqrt(S0.f32)
# --- compiled pseudocode ---
D0.f32 = sqrt(S0.f32)
# --- end pseudocode ---
def _VOP3Op_V_SQRT_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f64 = sqrt(S0.f64)
# --- compiled pseudocode ---
D0.f64 = sqrt(S0.f64)
# --- end pseudocode ---
def _VOP3Op_V_SIN_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = sin(S0.f32 * 32'F(PI * 2.0))
# --- compiled pseudocode ---
D0.f32 = sin(S0.f32 * F(PI * 2.0))
# --- end pseudocode ---
def _VOP3Op_V_COS_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = cos(S0.f32 * 32'F(PI * 2.0))
# --- compiled pseudocode ---
D0.f32 = cos(S0.f32 * F(PI * 2.0))
# --- end pseudocode ---
def _VOP3Op_V_NOT_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = ~S0.u32
# --- compiled pseudocode ---
D0.u32 = ~S0.u32
# --- end pseudocode ---
def _VOP3Op_V_BFREV_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32[31 : 0] = S0.u32[0 : 31]
# --- compiled pseudocode ---
D0.u32[31 : 0] = S0.u32[0 : 31]
# --- end pseudocode ---
def _VOP3Op_V_CLZ_I32_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = -1;
# // Set if no ones are found
# for i in 0 : 31 do
# // Search from MSB
# if S0.u32[31 - i] == 1'1U then
# D0.i32 = i;
# endif
# endfor
# --- compiled pseudocode ---
D0.i32 = -1
for i in range(0, int(31)+1):
if S0.u32[31 - i] == 1:
D0.i32 = i; break
# --- end pseudocode ---
def _VOP3Op_V_CTZ_I32_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = -1;
# // Set if no ones are found
# for i in 0 : 31 do
# // Search from LSB
# if S0.u32[i] == 1'1U then
# D0.i32 = i;
# endif
# endfor
# --- compiled pseudocode ---
D0.i32 = -1
for i in range(0, int(31)+1):
if S0.u32[i] == 1:
D0.i32 = i; break
# --- end pseudocode ---
def _VOP3Op_V_CLS_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = -1;
# // Set if all bits are the same
# for i in 1 : 31 do
# // Search from MSB
# if S0.i32[31 - i] != S0.i32[31] then
# D0.i32 = i;
# endif
# endfor
# --- compiled pseudocode ---
D0.i32 = -1
for i in range(1, int(31)+1):
if S0.i32[31 - i] != S0.i32[31]:
D0.i32 = i
# --- end pseudocode ---
def _VOP3Op_V_FREXP_EXP_I32_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if ((S0.f64 == +INF) || (S0.f64 == -INF) || isNAN(S0.f64)) then
# D0.i32 = 0
# else
# D0.i32 = exponent(S0.f64) - 1023 + 1
# endif
# --- compiled pseudocode ---
if ((S0.f64 == INF) or (S0.f64 == (-INF)) or isNAN(S0.f64)):
D0.i32 = 0
else:
D0.i32 = exponent(S0.f64) - 1023 + 1
# --- end pseudocode ---
def _VOP3Op_V_FREXP_MANT_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if ((S0.f64 == +INF) || (S0.f64 == -INF) || isNAN(S0.f64)) then
# D0.f64 = S0.f64
# else
# D0.f64 = mantissa(S0.f64)
# endif
# --- compiled pseudocode ---
if ((S0.f64 == INF) or (S0.f64 == (-INF)) or isNAN(S0.f64)):
D0.f64 = S0.f64
else:
D0.f64 = mantissa(S0.f64)
# --- end pseudocode ---
def _VOP3Op_V_FRACT_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f64 = S0.f64 + -floor(S0.f64)
# --- compiled pseudocode ---
D0.f64 = S0.f64 + -floor(S0.f64)
# --- end pseudocode ---
def _VOP3Op_V_FREXP_EXP_I32_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if ((64'F(S0.f32) == +INF) || (64'F(S0.f32) == -INF) || isNAN(64'F(S0.f32))) then
# D0.i32 = 0
# else
# D0.i32 = exponent(S0.f32) - 127 + 1
# endif
# --- compiled pseudocode ---
if ((F(S0.f32) == INF) or (F(S0.f32) == (-INF)) or isNAN(F(S0.f32))):
D0.i32 = 0
else:
D0.i32 = exponent(S0.f32) - 127 + 1
# --- end pseudocode ---
def _VOP3Op_V_FREXP_MANT_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if ((64'F(S0.f32) == +INF) || (64'F(S0.f32) == -INF) || isNAN(64'F(S0.f32))) then
# D0.f32 = S0.f32
# else
# D0.f32 = mantissa(S0.f32)
# endif
# --- compiled pseudocode ---
if ((F(S0.f32) == INF) or (F(S0.f32) == (-INF)) or isNAN(F(S0.f32))):
D0.f32 = S0.f32
else:
D0.f32 = mantissa(S0.f32)
# --- end pseudocode ---
def _VOP3Op_V_MOVRELS_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# addr = SRC0.u32;
# // Raw value from instruction
# D0.b32 = VGPR[laneId][addr].b32
# --- compiled pseudocode ---
addr = SRC0.u32
D0.b32 = VGPR[laneId][addr].b32
# --- end pseudocode ---
def _VOP3Op_V_CVT_F16_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = u16_to_f16(S0.u16)
# --- compiled pseudocode ---
D0.f16 = u16_to_f16(S0.u16)
# --- end pseudocode ---
def _VOP3Op_V_CVT_F16_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = i16_to_f16(S0.i16)
# --- compiled pseudocode ---
D0.f16 = i16_to_f16(S0.i16)
# --- end pseudocode ---
def _VOP3Op_V_CVT_U16_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u16 = f16_to_u16(S0.f16)
# --- compiled pseudocode ---
D0.u16 = f16_to_u16(S0.f16)
# --- end pseudocode ---
def _VOP3Op_V_CVT_I16_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i16 = f16_to_i16(S0.f16)
# --- compiled pseudocode ---
D0.i16 = f16_to_i16(S0.f16)
# --- end pseudocode ---
def _VOP3Op_V_RCP_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = 16'1.0 / S0.f16
# --- compiled pseudocode ---
D0.f16 = 1.0 / S0.f16
# --- end pseudocode ---
def _VOP3Op_V_SQRT_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = sqrt(S0.f16)
# --- compiled pseudocode ---
D0.f16 = sqrt(S0.f16)
# --- end pseudocode ---
def _VOP3Op_V_RSQ_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = 16'1.0 / sqrt(S0.f16)
# --- compiled pseudocode ---
D0.f16 = 1.0 / sqrt(S0.f16)
# --- end pseudocode ---
def _VOP3Op_V_LOG_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = log2(S0.f16)
# --- compiled pseudocode ---
D0.f16 = log2(S0.f16)
# --- end pseudocode ---
def _VOP3Op_V_EXP_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = pow(16'2.0, S0.f16)
# --- compiled pseudocode ---
D0.f16 = pow(2.0, S0.f16)
# --- end pseudocode ---
def _VOP3Op_V_FREXP_MANT_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if ((64'F(S0.f16) == +INF) || (64'F(S0.f16) == -INF) || isNAN(64'F(S0.f16))) then
# D0.f16 = S0.f16
# else
# D0.f16 = mantissa(S0.f16)
# endif
# --- compiled pseudocode ---
if ((F(S0.f16) == INF) or (F(S0.f16) == (-INF)) or isNAN(F(S0.f16))):
D0.f16 = S0.f16
else:
D0.f16 = mantissa(S0.f16)
# --- end pseudocode ---
def _VOP3Op_V_FREXP_EXP_I16_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if ((64'F(S0.f16) == +INF) || (64'F(S0.f16) == -INF) || isNAN(64'F(S0.f16))) then
# D0.i16 = 16'0
# else
# D0.i16 = 16'I(exponent(S0.f16) - 15 + 1)
# endif
# --- compiled pseudocode ---
if ((F(S0.f16) == INF) or (F(S0.f16) == (-INF)) or isNAN(F(S0.f16))):
D0.i16 = 0
else:
D0.i16 = (exponent(S0.f16) - 15 + 1)
# --- end pseudocode ---
def _VOP3Op_V_FLOOR_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = trunc(S0.f16);
# if ((S0.f16 < 16'0.0) && (S0.f16 != D0.f16)) then
# D0.f16 += -16'1.0
# endif
# --- compiled pseudocode ---
D0.f16 = trunc(S0.f16)
if ((S0.f16 < 0.0) and (S0.f16 != D0.f16)):
D0.f16 += -1.0
# --- end pseudocode ---
def _VOP3Op_V_CEIL_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = trunc(S0.f16);
# if ((S0.f16 > 16'0.0) && (S0.f16 != D0.f16)) then
# D0.f16 += 16'1.0
# endif
# --- compiled pseudocode ---
D0.f16 = trunc(S0.f16)
if ((S0.f16 > 0.0) and (S0.f16 != D0.f16)):
D0.f16 += 1.0
# --- end pseudocode ---
def _VOP3Op_V_TRUNC_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = trunc(S0.f16)
# --- compiled pseudocode ---
D0.f16 = trunc(S0.f16)
# --- end pseudocode ---
def _VOP3Op_V_RNDNE_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = floor(S0.f16 + 16'0.5);
# if (isEven(64'F(floor(S0.f16))) && (fract(S0.f16) == 16'0.5)) then
# D0.f16 -= 16'1.0
# endif
# --- compiled pseudocode ---
D0.f16 = floor(S0.f16 + 0.5)
if (isEven(F(floor(S0.f16))) and (fract(S0.f16) == 0.5)):
D0.f16 -= 1.0
# --- end pseudocode ---
def _VOP3Op_V_FRACT_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = S0.f16 + -floor(S0.f16)
# --- compiled pseudocode ---
D0.f16 = S0.f16 + -floor(S0.f16)
# --- end pseudocode ---
def _VOP3Op_V_SIN_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = sin(S0.f16 * 16'F(PI * 2.0))
# --- compiled pseudocode ---
D0.f16 = sin(S0.f16 * F(PI * 2.0))
# --- end pseudocode ---
def _VOP3Op_V_COS_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = cos(S0.f16 * 16'F(PI * 2.0))
# --- compiled pseudocode ---
D0.f16 = cos(S0.f16 * F(PI * 2.0))
# --- end pseudocode ---
def _VOP3Op_V_CVT_NORM_I16_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i16 = f16_to_snorm(S0.f16)
# --- compiled pseudocode ---
D0.i16 = f16_to_snorm(S0.f16)
# --- end pseudocode ---
def _VOP3Op_V_CVT_NORM_U16_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u16 = f16_to_unorm(S0.f16)
# --- compiled pseudocode ---
D0.u16 = f16_to_unorm(S0.f16)
# --- end pseudocode ---
def _VOP3Op_V_NOT_B16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u16 = ~S0.u16
# --- compiled pseudocode ---
D0.u16 = ~S0.u16
# --- end pseudocode ---
def _VOP3Op_V_CVT_I32_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = 32'I(signext(S0.i16))
# --- compiled pseudocode ---
D0.i32 = (signext(S0.i16))
# --- end pseudocode ---
def _VOP3Op_V_CVT_U32_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0 = { 16'0, S0.u16 }
# --- compiled pseudocode ---
D0.b32 = _pack(0, S0.u16)
# --- end pseudocode ---
def _VOP3Op_V_CNDMASK_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = VCC.u64[laneId] ? S1.u32 : S0.u32
# --- compiled pseudocode ---
D0.u32 = ((S1.u32) if (VCC.u64[laneId]) else (S0.u32))
# --- end pseudocode ---
def _VOP3Op_V_ADD_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = S0.f32 + S1.f32
# --- compiled pseudocode ---
D0.f32 = S0.f32 + S1.f32
# --- end pseudocode ---
def _VOP3Op_V_SUB_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = S0.f32 - S1.f32
# --- compiled pseudocode ---
D0.f32 = S0.f32 - S1.f32
# --- end pseudocode ---
def _VOP3Op_V_SUBREV_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = S1.f32 - S0.f32
# --- compiled pseudocode ---
D0.f32 = S1.f32 - S0.f32
# --- end pseudocode ---
def _VOP3Op_V_FMAC_DX9_ZERO_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if ((64'F(S0.f32) == 0.0) || (64'F(S1.f32) == 0.0)) then
# // DX9 rules, 0.0 * x = 0.0
# D0.f32 = S2.f32
# else
# D0.f32 = fma(S0.f32, S1.f32, D0.f32)
# endif
# --- compiled pseudocode ---
if ((F(S0.f32) == 0.0) or (F(S1.f32) == 0.0)):
D0.f32 = S2.f32
else:
D0.f32 = fma(S0.f32, S1.f32, D0.f32)
# --- end pseudocode ---
def _VOP3Op_V_MUL_DX9_ZERO_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if ((64'F(S0.f32) == 0.0) || (64'F(S1.f32) == 0.0)) then
# // DX9 rules, 0.0 * x = 0.0
# D0.f32 = 0.0F
# else
# D0.f32 = S0.f32 * S1.f32
# endif
# --- compiled pseudocode ---
if ((F(S0.f32) == 0.0) or (F(S1.f32) == 0.0)):
D0.f32 = 0.0
else:
D0.f32 = S0.f32 * S1.f32
# --- end pseudocode ---
def _VOP3Op_V_MUL_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = S0.f32 * S1.f32
# --- compiled pseudocode ---
D0.f32 = S0.f32 * S1.f32
# --- end pseudocode ---
def _VOP3Op_V_MUL_I32_I24(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = 32'I(S0.i24) * 32'I(S1.i24)
# --- compiled pseudocode ---
D0.i32 = (S0.i24) * (S1.i24)
# --- end pseudocode ---
def _VOP3Op_V_MUL_HI_I32_I24(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = 32'I((64'I(S0.i24) * 64'I(S1.i24)) >> 32U)
# --- compiled pseudocode ---
D0.i32 = (((S0.i24) * (S1.i24)) >> 32)
# --- end pseudocode ---
def _VOP3Op_V_MUL_U32_U24(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = 32'U(S0.u24) * 32'U(S1.u24)
# --- compiled pseudocode ---
D0.u32 = (S0.u24) * (S1.u24)
# --- end pseudocode ---
def _VOP3Op_V_MUL_HI_U32_U24(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = 32'U((64'U(S0.u24) * 64'U(S1.u24)) >> 32U)
# --- compiled pseudocode ---
D0.u32 = (((S0.u24) * (S1.u24)) >> 32)
# --- end pseudocode ---
def _VOP3Op_V_MIN_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# // Version of comparison where -0.0 < +0.0, differs from IEEE
# if WAVE_MODE.IEEE then
# if isSignalNAN(64'F(S0.f32)) then
# D0.f32 = 32'F(cvtToQuietNAN(64'F(S0.f32)))
# elsif isSignalNAN(64'F(S1.f32)) then
# D0.f32 = 32'F(cvtToQuietNAN(64'F(S1.f32)))
# elsif isQuietNAN(64'F(S1.f32)) then
# D0.f32 = S0.f32
# elsif isQuietNAN(64'F(S0.f32)) then
# D0.f32 = S1.f32
# elsif LT_NEG_ZERO(S0.f32, S1.f32) then
# // NOTE: -0<+0 is TRUE in this comparison
# D0.f32 = S0.f32
# else
# D0.f32 = S1.f32
# endif
# else
# if isNAN(64'F(S1.f32)) then
# D0.f32 = S0.f32
# elsif isNAN(64'F(S0.f32)) then
# D0.f32 = S1.f32
# elsif LT_NEG_ZERO(S0.f32, S1.f32) then
# // NOTE: -0<+0 is TRUE in this comparison
# D0.f32 = S0.f32
# else
# D0.f32 = S1.f32
# endif
# endif;
# // Inequalities in the above pseudocode behave differently from IEEE
# --- compiled pseudocode ---
if WAVE_MODE.IEEE:
if isSignalNAN(F(S0.f32)):
D0.f32 = F(cvtToQuietNAN(F(S0.f32)))
elif isSignalNAN(F(S1.f32)):
D0.f32 = F(cvtToQuietNAN(F(S1.f32)))
elif isQuietNAN(F(S1.f32)):
D0.f32 = S0.f32
elif isQuietNAN(F(S0.f32)):
D0.f32 = S1.f32
elif LT_NEG_ZERO(S0.f32, S1.f32):
D0.f32 = S0.f32
else:
D0.f32 = S1.f32
else:
if isNAN(F(S1.f32)):
D0.f32 = S0.f32
elif isNAN(F(S0.f32)):
D0.f32 = S1.f32
elif LT_NEG_ZERO(S0.f32, S1.f32):
D0.f32 = S0.f32
else:
D0.f32 = S1.f32
# --- end pseudocode ---
def _VOP3Op_V_MAX_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# // Version of comparison where +0.0 > -0.0, differs from IEEE
# if WAVE_MODE.IEEE then
# if isSignalNAN(64'F(S0.f32)) then
# D0.f32 = 32'F(cvtToQuietNAN(64'F(S0.f32)))
# elsif isSignalNAN(64'F(S1.f32)) then
# D0.f32 = 32'F(cvtToQuietNAN(64'F(S1.f32)))
# elsif isQuietNAN(64'F(S1.f32)) then
# D0.f32 = S0.f32
# elsif isQuietNAN(64'F(S0.f32)) then
# D0.f32 = S1.f32
# elsif GT_NEG_ZERO(S0.f32, S1.f32) then
# // NOTE: +0>-0 is TRUE in this comparison
# D0.f32 = S0.f32
# else
# D0.f32 = S1.f32
# endif
# else
# if isNAN(64'F(S1.f32)) then
# D0.f32 = S0.f32
# elsif isNAN(64'F(S0.f32)) then
# D0.f32 = S1.f32
# elsif GT_NEG_ZERO(S0.f32, S1.f32) then
# // NOTE: +0>-0 is TRUE in this comparison
# D0.f32 = S0.f32
# else
# D0.f32 = S1.f32
# endif
# endif;
# // Inequalities in the above pseudocode behave differently from IEEE
# --- compiled pseudocode ---
if WAVE_MODE.IEEE:
if isSignalNAN(F(S0.f32)):
D0.f32 = F(cvtToQuietNAN(F(S0.f32)))
elif isSignalNAN(F(S1.f32)):
D0.f32 = F(cvtToQuietNAN(F(S1.f32)))
elif isQuietNAN(F(S1.f32)):
D0.f32 = S0.f32
elif isQuietNAN(F(S0.f32)):
D0.f32 = S1.f32
elif GT_NEG_ZERO(S0.f32, S1.f32):
D0.f32 = S0.f32
else:
D0.f32 = S1.f32
else:
if isNAN(F(S1.f32)):
D0.f32 = S0.f32
elif isNAN(F(S0.f32)):
D0.f32 = S1.f32
elif GT_NEG_ZERO(S0.f32, S1.f32):
D0.f32 = S0.f32
else:
D0.f32 = S1.f32
# --- end pseudocode ---
def _VOP3Op_V_MIN_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = S0.i32 < S1.i32 ? S0.i32 : S1.i32
# --- compiled pseudocode ---
D0.i32 = ((S0.i32) if (S0.i32 < S1.i32) else (S1.i32))
# --- end pseudocode ---
def _VOP3Op_V_MAX_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = S0.i32 >= S1.i32 ? S0.i32 : S1.i32
# --- compiled pseudocode ---
D0.i32 = ((S0.i32) if (S0.i32 >= S1.i32) else (S1.i32))
# --- end pseudocode ---
def _VOP3Op_V_MIN_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = S0.u32 < S1.u32 ? S0.u32 : S1.u32
# --- compiled pseudocode ---
D0.u32 = ((S0.u32) if (S0.u32 < S1.u32) else (S1.u32))
# --- end pseudocode ---
def _VOP3Op_V_MAX_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = S0.u32 >= S1.u32 ? S0.u32 : S1.u32
# --- compiled pseudocode ---
D0.u32 = ((S0.u32) if (S0.u32 >= S1.u32) else (S1.u32))
# --- end pseudocode ---
def _VOP3Op_V_LSHLREV_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = (S1.u32 << S0[4 : 0].u32)
# --- compiled pseudocode ---
D0.u32 = (S1.u32 << S0[4 : 0].u32)
# --- end pseudocode ---
def _VOP3Op_V_LSHRREV_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = (S1.u32 >> S0[4 : 0].u32)
# --- compiled pseudocode ---
D0.u32 = (S1.u32 >> S0[4 : 0].u32)
# --- end pseudocode ---
def _VOP3Op_V_ASHRREV_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = (S1.i32 >> S0[4 : 0].u32)
# --- compiled pseudocode ---
D0.i32 = (S1.i32 >> S0[4 : 0].u32)
# --- end pseudocode ---
def _VOP3Op_V_AND_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = (S0.u32 & S1.u32)
# --- compiled pseudocode ---
D0.u32 = (S0.u32 & S1.u32)
# --- end pseudocode ---
def _VOP3Op_V_OR_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = (S0.u32 | S1.u32)
# --- compiled pseudocode ---
D0.u32 = (S0.u32 | S1.u32)
# --- end pseudocode ---
def _VOP3Op_V_XOR_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = (S0.u32 ^ S1.u32)
# --- compiled pseudocode ---
D0.u32 = (S0.u32 ^ S1.u32)
# --- end pseudocode ---
def _VOP3Op_V_XNOR_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = ~(S0.u32 ^ S1.u32)
# --- compiled pseudocode ---
D0.u32 = ~(S0.u32 ^ S1.u32)
# --- end pseudocode ---
def _VOP3Op_V_ADD_NC_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = S0.u32 + S1.u32
# --- compiled pseudocode ---
D0.u32 = S0.u32 + S1.u32
# --- end pseudocode ---
def _VOP3Op_V_SUB_NC_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = S0.u32 - S1.u32
# --- compiled pseudocode ---
D0.u32 = S0.u32 - S1.u32
# --- end pseudocode ---
def _VOP3Op_V_SUBREV_NC_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = S1.u32 - S0.u32
# --- compiled pseudocode ---
D0.u32 = S1.u32 - S0.u32
# --- end pseudocode ---
def _VOP3Op_V_FMAC_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = fma(S0.f32, S1.f32, D0.f32)
# --- compiled pseudocode ---
D0.f32 = fma(S0.f32, S1.f32, D0.f32)
# --- end pseudocode ---
def _VOP3Op_V_CVT_PK_RTZ_F16_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# prev_mode = ROUND_MODE;
# tmp[15 : 0].f16 = f32_to_f16(S0.f32);
# tmp[31 : 16].f16 = f32_to_f16(S1.f32);
tmp = Reg(0)
# --- compiled pseudocode ---
prev_mode = ROUND_MODE
tmp[15 : 0].f16 = f32_to_f16(S0.f32)
tmp[31 : 16].f16 = f32_to_f16(S1.f32)
# --- end pseudocode ---
def _VOP3Op_V_ADD_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = S0.f16 + S1.f16
# --- compiled pseudocode ---
D0.f16 = S0.f16 + S1.f16
# --- end pseudocode ---
def _VOP3Op_V_SUB_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = S0.f16 - S1.f16
# --- compiled pseudocode ---
D0.f16 = S0.f16 - S1.f16
# --- end pseudocode ---
def _VOP3Op_V_SUBREV_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = S1.f16 - S0.f16
# --- compiled pseudocode ---
D0.f16 = S1.f16 - S0.f16
# --- end pseudocode ---
def _VOP3Op_V_MUL_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = S0.f16 * S1.f16
# --- compiled pseudocode ---
D0.f16 = S0.f16 * S1.f16
# --- end pseudocode ---
def _VOP3Op_V_FMAC_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = fma(S0.f16, S1.f16, D0.f16)
# --- compiled pseudocode ---
D0.f16 = fma(S0.f16, S1.f16, D0.f16)
# --- end pseudocode ---
def _VOP3Op_V_MAX_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# // Version of comparison where +0.0 > -0.0, differs from IEEE
# if WAVE_MODE.IEEE then
# if isSignalNAN(64'F(S0.f16)) then
# D0.f16 = 16'F(cvtToQuietNAN(64'F(S0.f16)))
# elsif isSignalNAN(64'F(S1.f16)) then
# D0.f16 = 16'F(cvtToQuietNAN(64'F(S1.f16)))
# elsif isQuietNAN(64'F(S1.f16)) then
# D0.f16 = S0.f16
# elsif isQuietNAN(64'F(S0.f16)) then
# D0.f16 = S1.f16
# elsif GT_NEG_ZERO(S0.f16, S1.f16) then
# // NOTE: +0>-0 is TRUE in this comparison
# D0.f16 = S0.f16
# else
# D0.f16 = S1.f16
# endif
# else
# if isNAN(64'F(S1.f16)) then
# D0.f16 = S0.f16
# elsif isNAN(64'F(S0.f16)) then
# D0.f16 = S1.f16
# elsif GT_NEG_ZERO(S0.f16, S1.f16) then
# // NOTE: +0>-0 is TRUE in this comparison
# D0.f16 = S0.f16
# else
# D0.f16 = S1.f16
# endif
# endif;
# // Inequalities in the above pseudocode behave differently from IEEE
# --- compiled pseudocode ---
if WAVE_MODE.IEEE:
if isSignalNAN(F(S0.f16)):
D0.f16 = F(cvtToQuietNAN(F(S0.f16)))
elif isSignalNAN(F(S1.f16)):
D0.f16 = F(cvtToQuietNAN(F(S1.f16)))
elif isQuietNAN(F(S1.f16)):
D0.f16 = S0.f16
elif isQuietNAN(F(S0.f16)):
D0.f16 = S1.f16
elif GT_NEG_ZERO(S0.f16, S1.f16):
D0.f16 = S0.f16
else:
D0.f16 = S1.f16
else:
if isNAN(F(S1.f16)):
D0.f16 = S0.f16
elif isNAN(F(S0.f16)):
D0.f16 = S1.f16
elif GT_NEG_ZERO(S0.f16, S1.f16):
D0.f16 = S0.f16
else:
D0.f16 = S1.f16
# --- end pseudocode ---
def _VOP3Op_V_MIN_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# // Version of comparison where -0.0 < +0.0, differs from IEEE
# if WAVE_MODE.IEEE then
# if isSignalNAN(64'F(S0.f16)) then
# D0.f16 = 16'F(cvtToQuietNAN(64'F(S0.f16)))
# elsif isSignalNAN(64'F(S1.f16)) then
# D0.f16 = 16'F(cvtToQuietNAN(64'F(S1.f16)))
# elsif isQuietNAN(64'F(S1.f16)) then
# D0.f16 = S0.f16
# elsif isQuietNAN(64'F(S0.f16)) then
# D0.f16 = S1.f16
# elsif LT_NEG_ZERO(S0.f16, S1.f16) then
# // NOTE: -0<+0 is TRUE in this comparison
# D0.f16 = S0.f16
# else
# D0.f16 = S1.f16
# endif
# else
# if isNAN(64'F(S1.f16)) then
# D0.f16 = S0.f16
# elsif isNAN(64'F(S0.f16)) then
# D0.f16 = S1.f16
# elsif LT_NEG_ZERO(S0.f16, S1.f16) then
# // NOTE: -0<+0 is TRUE in this comparison
# D0.f16 = S0.f16
# else
# D0.f16 = S1.f16
# endif
# endif;
# // Inequalities in the above pseudocode behave differently from IEEE
# --- compiled pseudocode ---
if WAVE_MODE.IEEE:
if isSignalNAN(F(S0.f16)):
D0.f16 = F(cvtToQuietNAN(F(S0.f16)))
elif isSignalNAN(F(S1.f16)):
D0.f16 = F(cvtToQuietNAN(F(S1.f16)))
elif isQuietNAN(F(S1.f16)):
D0.f16 = S0.f16
elif isQuietNAN(F(S0.f16)):
D0.f16 = S1.f16
elif LT_NEG_ZERO(S0.f16, S1.f16):
D0.f16 = S0.f16
else:
D0.f16 = S1.f16
else:
if isNAN(F(S1.f16)):
D0.f16 = S0.f16
elif isNAN(F(S0.f16)):
D0.f16 = S1.f16
elif LT_NEG_ZERO(S0.f16, S1.f16):
D0.f16 = S0.f16
else:
D0.f16 = S1.f16
# --- end pseudocode ---
def _VOP3Op_V_LDEXP_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = S0.f16 * 16'F(2.0F ** 32'I(S1.i16))
# --- compiled pseudocode ---
D0.f16 = S0.f16 * F(2.0 ** (S1.i16))
# --- end pseudocode ---
def _VOP3Op_V_FMA_DX9_ZERO_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if ((64'F(S0.f32) == 0.0) || (64'F(S1.f32) == 0.0)) then
# // DX9 rules, 0.0 * x = 0.0
# D0.f32 = S2.f32
# else
# D0.f32 = fma(S0.f32, S1.f32, S2.f32)
# endif
# --- compiled pseudocode ---
if ((F(S0.f32) == 0.0) or (F(S1.f32) == 0.0)):
D0.f32 = S2.f32
else:
D0.f32 = fma(S0.f32, S1.f32, S2.f32)
# --- end pseudocode ---
def _VOP3Op_V_MAD_I32_I24(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = 32'I(S0.i24) * 32'I(S1.i24) + S2.i32
# --- compiled pseudocode ---
D0.i32 = (S0.i24) * (S1.i24) + S2.i32
# --- end pseudocode ---
def _VOP3Op_V_MAD_U32_U24(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = 32'U(S0.u24) * 32'U(S1.u24) + S2.u32
# --- compiled pseudocode ---
D0.u32 = (S0.u24) * (S1.u24) + S2.u32
# --- end pseudocode ---
def _VOP3Op_V_CUBEID_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# // Set D0.f = cubemap face ID ({0.0, 1.0, ..., 5.0}).
# // XYZ coordinate is given in (S0.f, S1.f, S2.f).
# // S0.f = x
# // S1.f = y
# // S2.f = z
# if ((abs(S2.f32) >= abs(S0.f32)) && (abs(S2.f32) >= abs(S1.f32))) then
# if S2.f32 < 0.0F then
# D0.f32 = 5.0F
# else
# D0.f32 = 4.0F
# endif
# elsif abs(S1.f32) >= abs(S0.f32) then
# if S1.f32 < 0.0F then
# D0.f32 = 3.0F
# else
# D0.f32 = 2.0F
# endif
# else
# if S0.f32 < 0.0F then
# D0.f32 = 1.0F
# else
# D0.f32 = 0.0F
# endif
# endif
# --- compiled pseudocode ---
if ((abs(S2.f32) >= abs(S0.f32)) and (abs(S2.f32) >= abs(S1.f32))):
if S2.f32 < 0.0:
D0.f32 = 5.0
else:
D0.f32 = 4.0
elif abs(S1.f32) >= abs(S0.f32):
if S1.f32 < 0.0:
D0.f32 = 3.0
else:
D0.f32 = 2.0
else:
if S0.f32 < 0.0:
D0.f32 = 1.0
else:
D0.f32 = 0.0
# --- end pseudocode ---
def _VOP3Op_V_CUBESC_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# // D0.f = cubemap S coordinate.
# // XYZ coordinate is given in (S0.f, S1.f, S2.f).
# // S0.f = x
# // S1.f = y
# // S2.f = z
# if ((abs(S2.f32) >= abs(S0.f32)) && (abs(S2.f32) >= abs(S1.f32))) then
# if S2.f32 < 0.0F then
# D0.f32 = -S0.f32
# else
# D0.f32 = S0.f32
# endif
# elsif abs(S1.f32) >= abs(S0.f32) then
# D0.f32 = S0.f32
# else
# if S0.f32 < 0.0F then
# D0.f32 = S2.f32
# else
# D0.f32 = -S2.f32
# endif
# endif
# --- compiled pseudocode ---
if ((abs(S2.f32) >= abs(S0.f32)) and (abs(S2.f32) >= abs(S1.f32))):
if S2.f32 < 0.0:
D0.f32 = -S0.f32
else:
D0.f32 = S0.f32
elif abs(S1.f32) >= abs(S0.f32):
D0.f32 = S0.f32
else:
if S0.f32 < 0.0:
D0.f32 = S2.f32
else:
D0.f32 = -S2.f32
# --- end pseudocode ---
def _VOP3Op_V_CUBETC_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# // D0.f = cubemap T coordinate.
# // XYZ coordinate is given in (S0.f, S1.f, S2.f).
# // S0.f = x
# // S1.f = y
# // S2.f = z
# if ((abs(S2.f32) >= abs(S0.f32)) && (abs(S2.f32) >= abs(S1.f32))) then
# D0.f32 = -S1.f32
# elsif abs(S1.f32) >= abs(S0.f32) then
# if S1.f32 < 0.0F then
# D0.f32 = -S2.f32
# else
# D0.f32 = S2.f32
# endif
# else
# D0.f32 = -S1.f32
# endif
# --- compiled pseudocode ---
if ((abs(S2.f32) >= abs(S0.f32)) and (abs(S2.f32) >= abs(S1.f32))):
D0.f32 = -S1.f32
elif abs(S1.f32) >= abs(S0.f32):
if S1.f32 < 0.0:
D0.f32 = -S2.f32
else:
D0.f32 = S2.f32
else:
D0.f32 = -S1.f32
# --- end pseudocode ---
def _VOP3Op_V_CUBEMA_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# // D0.f = 2.0 * cubemap major axis.
# // XYZ coordinate is given in (S0.f, S1.f, S2.f).
# // S0.f = x
# // S1.f = y
# // S2.f = z
# if ((abs(S2.f32) >= abs(S0.f32)) && (abs(S2.f32) >= abs(S1.f32))) then
# D0.f32 = S2.f32 * 2.0F
# elsif abs(S1.f32) >= abs(S0.f32) then
# D0.f32 = S1.f32 * 2.0F
# else
# D0.f32 = S0.f32 * 2.0F
# endif
# --- compiled pseudocode ---
if ((abs(S2.f32) >= abs(S0.f32)) and (abs(S2.f32) >= abs(S1.f32))):
D0.f32 = S2.f32 * 2.0
elif abs(S1.f32) >= abs(S0.f32):
D0.f32 = S1.f32 * 2.0
else:
D0.f32 = S0.f32 * 2.0
# --- end pseudocode ---
def _VOP3Op_V_BFE_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = ((S0.u32 >> S1[4 : 0].u32) & ((1U << S2[4 : 0].u32) - 1U))
# --- compiled pseudocode ---
D0.u32 = ((S0.u32 >> S1[4 : 0].u32) & ((1 << S2[4 : 0].u32) - 1))
# --- end pseudocode ---
def _VOP3Op_V_BFE_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp.i32 = ((S0.i32 >> S1[4 : 0].u32) & ((1 << S2[4 : 0].u32) - 1));
# D0.i32 = signext_from_bit(tmp.i32, S2[4 : 0].u32)
tmp = Reg(0)
# --- compiled pseudocode ---
tmp.i32 = ((S0.i32 >> S1[4 : 0].u32) & ((1 << S2[4 : 0].u32) - 1))
D0.i32 = signext_from_bit(tmp.i32, S2[4 : 0].u32)
# --- end pseudocode ---
def _VOP3Op_V_BFI_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = ((S0.u32 & S1.u32) | (~S0.u32 & S2.u32))
# --- compiled pseudocode ---
D0.u32 = ((S0.u32 & S1.u32) | (~S0.u32 & S2.u32))
# --- end pseudocode ---
def _VOP3Op_V_FMA_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = fma(S0.f32, S1.f32, S2.f32)
# --- compiled pseudocode ---
D0.f32 = fma(S0.f32, S1.f32, S2.f32)
# --- end pseudocode ---
def _VOP3Op_V_FMA_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f64 = fma(S0.f64, S1.f64, S2.f64)
# --- compiled pseudocode ---
D0.f64 = fma(S0.f64, S1.f64, S2.f64)
# --- end pseudocode ---
def _VOP3Op_V_LERP_U8(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = ((S0.u32[31 : 24] + S1.u32[31 : 24] + S2.u32[24].u8) >> 1U << 24U);
# tmp += ((S0.u32[23 : 16] + S1.u32[23 : 16] + S2.u32[16].u8) >> 1U << 16U);
# tmp += ((S0.u32[15 : 8] + S1.u32[15 : 8] + S2.u32[8].u8) >> 1U << 8U);
# tmp += ((S0.u32[7 : 0] + S1.u32[7 : 0] + S2.u32[0].u8) >> 1U);
# D0.u32 = tmp.u32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(((S0.u32[31 : 24] + S1.u32[31 : 24] + S2.u32[24].u8) >> 1 << 24))
tmp += ((S0.u32[23 : 16] + S1.u32[23 : 16] + S2.u32[16].u8) >> 1 << 16)
tmp += ((S0.u32[15 : 8] + S1.u32[15 : 8] + S2.u32[8].u8) >> 1 << 8)
tmp += ((S0.u32[7 : 0] + S1.u32[7 : 0] + S2.u32[0].u8) >> 1)
D0.u32 = tmp.u32
# --- end pseudocode ---
def _VOP3Op_V_ALIGNBIT_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = 32'U(({ S0.u32, S1.u32 } >> S2.u32[4 : 0].u32) & 0xffffffffLL)
# --- compiled pseudocode ---
D0.u32 = ((_pack32(S0.u32, S1.u32) >> S2.u32[4 : 0].u32) & 0xffffffff)
# --- end pseudocode ---
def _VOP3Op_V_ALIGNBYTE_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = 32'U(({ S0.u32, S1.u32 } >> (S2.u32[1 : 0].u32 * 8U)) & 0xffffffffLL)
# --- compiled pseudocode ---
D0.u32 = ((_pack32(S0.u32, S1.u32) >> (S2.u32[1 : 0].u32 * 8)) & 0xffffffff)
# --- end pseudocode ---
def _VOP3Op_V_MULLIT_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if ((S1.f32 == -MAX_FLOAT_F32) || (64'F(S1.f32) == -INF) || isNAN(64'F(S1.f32)) || (S2.f32 <= 0.0F) ||
# isNAN(64'F(S2.f32))) then
# D0.f32 = -MAX_FLOAT_F32
# else
# D0.f32 = S0.f32 * S1.f32
# endif
# --- compiled pseudocode ---
if ((S1.f32 == -MAX_FLOAT_F32) or (F(S1.f32) == (-INF)) or isNAN(F(S1.f32)) or (S2.f32 <= 0.0) or isNAN(F(S2.f32))):
D0.f32 = -MAX_FLOAT_F32
else:
D0.f32 = S0.f32 * S1.f32
# --- end pseudocode ---
def _VOP3Op_V_MIN3_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = v_min_f32(v_min_f32(S0.f32, S1.f32), S2.f32)
# --- compiled pseudocode ---
D0.f32 = v_min_f32(v_min_f32(S0.f32, S1.f32), S2.f32)
# --- end pseudocode ---
def _VOP3Op_V_MIN3_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = v_min_i32(v_min_i32(S0.i32, S1.i32), S2.i32)
# --- compiled pseudocode ---
D0.i32 = v_min_i32(v_min_i32(S0.i32, S1.i32), S2.i32)
# --- end pseudocode ---
def _VOP3Op_V_MIN3_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = v_min_u32(v_min_u32(S0.u32, S1.u32), S2.u32)
# --- compiled pseudocode ---
D0.u32 = v_min_u32(v_min_u32(S0.u32, S1.u32), S2.u32)
# --- end pseudocode ---
def _VOP3Op_V_MAX3_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = v_max_f32(v_max_f32(S0.f32, S1.f32), S2.f32)
# --- compiled pseudocode ---
D0.f32 = v_max_f32(v_max_f32(S0.f32, S1.f32), S2.f32)
# --- end pseudocode ---
def _VOP3Op_V_MAX3_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = v_max_i32(v_max_i32(S0.i32, S1.i32), S2.i32)
# --- compiled pseudocode ---
D0.i32 = v_max_i32(v_max_i32(S0.i32, S1.i32), S2.i32)
# --- end pseudocode ---
def _VOP3Op_V_MAX3_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = v_max_u32(v_max_u32(S0.u32, S1.u32), S2.u32)
# --- compiled pseudocode ---
D0.u32 = v_max_u32(v_max_u32(S0.u32, S1.u32), S2.u32)
# --- end pseudocode ---
def _VOP3Op_V_MED3_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if (isNAN(64'F(S0.f32)) || isNAN(64'F(S1.f32)) || isNAN(64'F(S2.f32))) then
# D0.f32 = v_min3_f32(S0.f32, S1.f32, S2.f32)
# elsif v_max3_f32(S0.f32, S1.f32, S2.f32) == S0.f32 then
# D0.f32 = v_max_f32(S1.f32, S2.f32)
# elsif v_max3_f32(S0.f32, S1.f32, S2.f32) == S1.f32 then
# D0.f32 = v_max_f32(S0.f32, S2.f32)
# else
# D0.f32 = v_max_f32(S0.f32, S1.f32)
# endif
# --- compiled pseudocode ---
if (isNAN(F(S0.f32)) or isNAN(F(S1.f32)) or isNAN(F(S2.f32))):
D0.f32 = v_min3_f32(S0.f32, S1.f32, S2.f32)
elif v_max3_f32(S0.f32, S1.f32, S2.f32) == S0.f32:
D0.f32 = v_max_f32(S1.f32, S2.f32)
elif v_max3_f32(S0.f32, S1.f32, S2.f32) == S1.f32:
D0.f32 = v_max_f32(S0.f32, S2.f32)
else:
D0.f32 = v_max_f32(S0.f32, S1.f32)
# --- end pseudocode ---
def _VOP3Op_V_MED3_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if v_max3_i32(S0.i32, S1.i32, S2.i32) == S0.i32 then
# D0.i32 = v_max_i32(S1.i32, S2.i32)
# elsif v_max3_i32(S0.i32, S1.i32, S2.i32) == S1.i32 then
# D0.i32 = v_max_i32(S0.i32, S2.i32)
# else
# D0.i32 = v_max_i32(S0.i32, S1.i32)
# endif
# --- compiled pseudocode ---
if v_max3_i32(S0.i32, S1.i32, S2.i32) == S0.i32:
D0.i32 = v_max_i32(S1.i32, S2.i32)
elif v_max3_i32(S0.i32, S1.i32, S2.i32) == S1.i32:
D0.i32 = v_max_i32(S0.i32, S2.i32)
else:
D0.i32 = v_max_i32(S0.i32, S1.i32)
# --- end pseudocode ---
def _VOP3Op_V_MED3_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if v_max3_u32(S0.u32, S1.u32, S2.u32) == S0.u32 then
# D0.u32 = v_max_u32(S1.u32, S2.u32)
# elsif v_max3_u32(S0.u32, S1.u32, S2.u32) == S1.u32 then
# D0.u32 = v_max_u32(S0.u32, S2.u32)
# else
# D0.u32 = v_max_u32(S0.u32, S1.u32)
# endif
# --- compiled pseudocode ---
if v_max3_u32(S0.u32, S1.u32, S2.u32) == S0.u32:
D0.u32 = v_max_u32(S1.u32, S2.u32)
elif v_max3_u32(S0.u32, S1.u32, S2.u32) == S1.u32:
D0.u32 = v_max_u32(S0.u32, S2.u32)
else:
D0.u32 = v_max_u32(S0.u32, S1.u32)
# --- end pseudocode ---
def _VOP3Op_V_SAD_U8(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# // UNSIGNED comparison
# tmp = S2.u32;
# tmp += 32'U(ABSDIFF(S0.u32[7 : 0], S1.u32[7 : 0]));
# tmp += 32'U(ABSDIFF(S0.u32[15 : 8], S1.u32[15 : 8]));
# tmp += 32'U(ABSDIFF(S0.u32[23 : 16], S1.u32[23 : 16]));
# tmp += 32'U(ABSDIFF(S0.u32[31 : 24], S1.u32[31 : 24]));
# D0.u32 = tmp
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(S2.u32)
tmp += (ABSDIFF(S0.u32[7 : 0], S1.u32[7 : 0]))
tmp += (ABSDIFF(S0.u32[15 : 8], S1.u32[15 : 8]))
tmp += (ABSDIFF(S0.u32[23 : 16], S1.u32[23 : 16]))
tmp += (ABSDIFF(S0.u32[31 : 24], S1.u32[31 : 24]))
D0.u32 = tmp
# --- end pseudocode ---
def _VOP3Op_V_SAD_HI_U8(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = (32'U(v_sad_u8(S0, S1, 0U)) << 16U) + S2.u32
# --- compiled pseudocode ---
D0.u32 = ((v_sad_u8(S0, S1, 0)) << 16) + S2.u32
# --- end pseudocode ---
def _VOP3Op_V_SAD_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# // UNSIGNED comparison
# tmp = S2.u32;
# tmp += ABSDIFF(S0[15 : 0].u16, S1[15 : 0].u16);
# tmp += ABSDIFF(S0[31 : 16].u16, S1[31 : 16].u16);
# D0.u32 = tmp
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(S2.u32)
tmp += ABSDIFF(S0[15 : 0].u16, S1[15 : 0].u16)
tmp += ABSDIFF(S0[31 : 16].u16, S1[31 : 16].u16)
D0.u32 = tmp
# --- end pseudocode ---
def _VOP3Op_V_SAD_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# // UNSIGNED comparison
# D0.u32 = ABSDIFF(S0.u32, S1.u32) + S2.u32
# --- compiled pseudocode ---
D0.u32 = ABSDIFF(S0.u32, S1.u32) + S2.u32
# --- end pseudocode ---
def _VOP3Op_V_CVT_PK_U8_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = (S2.u32 & 32'U(~(0xff << (S1.u32[1 : 0].u32 * 8U))));
# tmp = (tmp | ((32'U(f32_to_u8(S0.f32)) & 255U) << (S1.u32[1 : 0].u32 * 8U)));
# D0.u32 = tmp
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg((S2.u32 & (~(0xff << (S1.u32[1 : 0].u32 * 8)))))
tmp = Reg((tmp | (((f32_to_u8(S0.f32)) & 255) << (S1.u32[1 : 0].u32 * 8))))
D0.u32 = tmp
# --- end pseudocode ---
def _VOP3Op_V_DIV_FIXUP_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# sign_out = (sign(S1.f32) ^ sign(S2.f32));
# if isNAN(64'F(S2.f32)) then
# D0.f32 = 32'F(cvtToQuietNAN(64'F(S2.f32)))
# elsif isNAN(64'F(S1.f32)) then
# D0.f32 = 32'F(cvtToQuietNAN(64'F(S1.f32)))
# elsif ((64'F(S1.f32) == 0.0) && (64'F(S2.f32) == 0.0)) then
# // 0/0
# D0.f32 = 32'F(0xffc00000)
# elsif ((64'F(abs(S1.f32)) == +INF) && (64'F(abs(S2.f32)) == +INF)) then
# // inf/inf
# D0.f32 = 32'F(0xffc00000)
# elsif ((64'F(S1.f32) == 0.0) || (64'F(abs(S2.f32)) == +INF)) then
# // x/0, or inf/y
# D0.f32 = sign_out ? -INF.f32 : +INF.f32
# elsif ((64'F(abs(S1.f32)) == +INF) || (64'F(S2.f32) == 0.0)) then
# // x/inf, 0/y
# D0.f32 = sign_out ? -0.0F : 0.0F
# elsif exponent(S2.f32) - exponent(S1.f32) < -150 then
# D0.f32 = sign_out ? -UNDERFLOW_F32 : UNDERFLOW_F32
# elsif exponent(S1.f32) == 255 then
# D0.f32 = sign_out ? -OVERFLOW_F32 : OVERFLOW_F32
# else
# D0.f32 = sign_out ? -abs(S0.f32) : abs(S0.f32)
# endif
# --- compiled pseudocode ---
sign_out = (sign(S1.f32) ^ sign(S2.f32))
if isNAN(F(S2.f32)):
D0.f32 = F(cvtToQuietNAN(F(S2.f32)))
elif isNAN(F(S1.f32)):
D0.f32 = F(cvtToQuietNAN(F(S1.f32)))
elif ((F(S1.f32) == 0.0) and (F(S2.f32) == 0.0)):
D0.f32 = F(0xffc00000)
elif ((F(abs(S1.f32)) == INF) and (F(abs(S2.f32)) == INF)):
D0.f32 = F(0xffc00000)
elif ((F(S1.f32) == 0.0) or (F(abs(S2.f32)) == INF)):
D0.f32 = (((-INF).f32) if (sign_out) else (INF.f32))
elif ((F(abs(S1.f32)) == INF) or (F(S2.f32) == 0.0)):
D0.f32 = ((-0.0) if (sign_out) else (0.0))
elif exponent(S2.f32) - exponent(S1.f32) < -150:
D0.f32 = ((-UNDERFLOW_F32) if (sign_out) else (UNDERFLOW_F32))
elif exponent(S1.f32) == 255:
D0.f32 = ((-OVERFLOW_F32) if (sign_out) else (OVERFLOW_F32))
else:
D0.f32 = ((-abs(S0.f32)) if (sign_out) else (abs(S0.f32)))
# --- end pseudocode ---
def _VOP3Op_V_DIV_FIXUP_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# sign_out = (sign(S1.f64) ^ sign(S2.f64));
# if isNAN(S2.f64) then
# D0.f64 = cvtToQuietNAN(S2.f64)
# elsif isNAN(S1.f64) then
# D0.f64 = cvtToQuietNAN(S1.f64)
# elsif ((S1.f64 == 0.0) && (S2.f64 == 0.0)) then
# // 0/0
# D0.f64 = 64'F(0xfff8000000000000LL)
# elsif ((abs(S1.f64) == +INF) && (abs(S2.f64) == +INF)) then
# // inf/inf
# D0.f64 = 64'F(0xfff8000000000000LL)
# elsif ((S1.f64 == 0.0) || (abs(S2.f64) == +INF)) then
# // x/0, or inf/y
# D0.f64 = sign_out ? -INF : +INF
# elsif ((abs(S1.f64) == +INF) || (S2.f64 == 0.0)) then
# // x/inf, 0/y
# D0.f64 = sign_out ? -0.0 : 0.0
# elsif exponent(S2.f64) - exponent(S1.f64) < -1075 then
# D0.f64 = sign_out ? -UNDERFLOW_F64 : UNDERFLOW_F64
# elsif exponent(S1.f64) == 2047 then
# D0.f64 = sign_out ? -OVERFLOW_F64 : OVERFLOW_F64
# else
# D0.f64 = sign_out ? -abs(S0.f64) : abs(S0.f64)
# endif
# --- compiled pseudocode ---
sign_out = (sign(S1.f64) ^ sign(S2.f64))
if isNAN(S2.f64):
D0.f64 = cvtToQuietNAN(S2.f64)
elif isNAN(S1.f64):
D0.f64 = cvtToQuietNAN(S1.f64)
elif ((S1.f64 == 0.0) and (S2.f64 == 0.0)):
D0.f64 = F(0xfff8000000000000)
elif ((abs(S1.f64) == INF) and (abs(S2.f64) == INF)):
D0.f64 = F(0xfff8000000000000)
elif ((S1.f64 == 0.0) or (abs(S2.f64) == INF)):
D0.f64 = (((-INF)) if (sign_out) else (INF))
elif ((abs(S1.f64) == INF) or (S2.f64 == 0.0)):
D0.f64 = ((-0.0) if (sign_out) else (0.0))
elif exponent(S2.f64) - exponent(S1.f64) < -1075:
D0.f64 = ((-UNDERFLOW_F64) if (sign_out) else (UNDERFLOW_F64))
elif exponent(S1.f64) == 2047:
D0.f64 = ((-OVERFLOW_F64) if (sign_out) else (OVERFLOW_F64))
else:
D0.f64 = ((-abs(S0.f64)) if (sign_out) else (abs(S0.f64)))
# --- end pseudocode ---
def _VOP3Op_V_DIV_FMAS_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if VCC.u64[laneId] then
# D0.f32 = 2.0F ** 32 * fma(S0.f32, S1.f32, S2.f32)
# else
# D0.f32 = fma(S0.f32, S1.f32, S2.f32)
# endif
# --- compiled pseudocode ---
if VCC.u64[laneId]:
D0.f32 = 2.0 ** 32 * fma(S0.f32, S1.f32, S2.f32)
else:
D0.f32 = fma(S0.f32, S1.f32, S2.f32)
# --- end pseudocode ---
def _VOP3Op_V_DIV_FMAS_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if VCC.u64[laneId] then
# D0.f64 = 2.0 ** 64 * fma(S0.f64, S1.f64, S2.f64)
# else
# D0.f64 = fma(S0.f64, S1.f64, S2.f64)
# endif
# --- compiled pseudocode ---
if VCC.u64[laneId]:
D0.f64 = 2.0 ** 64 * fma(S0.f64, S1.f64, S2.f64)
else:
D0.f64 = fma(S0.f64, S1.f64, S2.f64)
# --- end pseudocode ---
def _VOP3Op_V_MSAD_U8(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# // UNSIGNED comparison
# tmp = S2.u32;
# tmp += S1.u32[7 : 0] == 8'0U ? 0U : 32'U(ABSDIFF(S0.u32[7 : 0], S1.u32[7 : 0]));
# tmp += S1.u32[15 : 8] == 8'0U ? 0U : 32'U(ABSDIFF(S0.u32[15 : 8], S1.u32[15 : 8]));
# tmp += S1.u32[23 : 16] == 8'0U ? 0U : 32'U(ABSDIFF(S0.u32[23 : 16], S1.u32[23 : 16]));
# tmp += S1.u32[31 : 24] == 8'0U ? 0U : 32'U(ABSDIFF(S0.u32[31 : 24], S1.u32[31 : 24]));
# D0.u32 = tmp
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(S2.u32)
tmp += ((0) if (S1.u32[7 : 0] == 0) else ((ABSDIFF(S0.u32[7 : 0], S1.u32[7 : 0]))))
tmp += ((0) if (S1.u32[15 : 8] == 0) else ((ABSDIFF(S0.u32[15 : 8], S1.u32[15 : 8]))))
tmp += ((0) if (S1.u32[23 : 16] == 0) else ((ABSDIFF(S0.u32[23 : 16], S1.u32[23 : 16]))))
tmp += ((0) if (S1.u32[31 : 24] == 0) else ((ABSDIFF(S0.u32[31 : 24], S1.u32[31 : 24]))))
D0.u32 = tmp
# --- end pseudocode ---
def _VOP3Op_V_QSAD_PK_U16_U8(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp[63 : 48] = 16'B(v_sad_u8(S0[55 : 24], S1[31 : 0], S2[63 : 48].u32));
# tmp[47 : 32] = 16'B(v_sad_u8(S0[47 : 16], S1[31 : 0], S2[47 : 32].u32));
# tmp[31 : 16] = 16'B(v_sad_u8(S0[39 : 8], S1[31 : 0], S2[31 : 16].u32));
# tmp[15 : 0] = 16'B(v_sad_u8(S0[31 : 0], S1[31 : 0], S2[15 : 0].u32));
# D0.b64 = tmp.b64
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[63 : 48] = (v_sad_u8(S0[55 : 24], S1[31 : 0], S2[63 : 48].u32))
tmp[47 : 32] = (v_sad_u8(S0[47 : 16], S1[31 : 0], S2[47 : 32].u32))
tmp[31 : 16] = (v_sad_u8(S0[39 : 8], S1[31 : 0], S2[31 : 16].u32))
tmp[15 : 0] = (v_sad_u8(S0[31 : 0], S1[31 : 0], S2[15 : 0].u32))
D0.b64 = tmp.b64
# --- end pseudocode ---
def _VOP3Op_V_MQSAD_PK_U16_U8(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp[63 : 48] = 16'B(v_msad_u8(S0[55 : 24], S1[31 : 0], S2[63 : 48].u32));
# tmp[47 : 32] = 16'B(v_msad_u8(S0[47 : 16], S1[31 : 0], S2[47 : 32].u32));
# tmp[31 : 16] = 16'B(v_msad_u8(S0[39 : 8], S1[31 : 0], S2[31 : 16].u32));
# tmp[15 : 0] = 16'B(v_msad_u8(S0[31 : 0], S1[31 : 0], S2[15 : 0].u32));
# D0.b64 = tmp.b64
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[63 : 48] = (v_msad_u8(S0[55 : 24], S1[31 : 0], S2[63 : 48].u32))
tmp[47 : 32] = (v_msad_u8(S0[47 : 16], S1[31 : 0], S2[47 : 32].u32))
tmp[31 : 16] = (v_msad_u8(S0[39 : 8], S1[31 : 0], S2[31 : 16].u32))
tmp[15 : 0] = (v_msad_u8(S0[31 : 0], S1[31 : 0], S2[15 : 0].u32))
D0.b64 = tmp.b64
# --- end pseudocode ---
def _VOP3Op_V_MQSAD_U32_U8(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp[127 : 96] = 32'B(v_msad_u8(S0[55 : 24], S1[31 : 0], S2[127 : 96].u32));
# tmp[95 : 64] = 32'B(v_msad_u8(S0[47 : 16], S1[31 : 0], S2[95 : 64].u32));
# tmp[63 : 32] = 32'B(v_msad_u8(S0[39 : 8], S1[31 : 0], S2[63 : 32].u32));
# tmp[31 : 0] = 32'B(v_msad_u8(S0[31 : 0], S1[31 : 0], S2[31 : 0].u32));
# D0.b128 = tmp.b128
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[127 : 96] = (v_msad_u8(S0[55 : 24], S1[31 : 0], S2[127 : 96].u32))
tmp[95 : 64] = (v_msad_u8(S0[47 : 16], S1[31 : 0], S2[95 : 64].u32))
tmp[63 : 32] = (v_msad_u8(S0[39 : 8], S1[31 : 0], S2[63 : 32].u32))
tmp[31 : 0] = (v_msad_u8(S0[31 : 0], S1[31 : 0], S2[31 : 0].u32))
D0.b128 = tmp.b128
# --- end pseudocode ---
def _VOP3Op_V_XOR3_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = (S0.u32 ^ S1.u32 ^ S2.u32)
# --- compiled pseudocode ---
D0.u32 = (S0.u32 ^ S1.u32 ^ S2.u32)
# --- end pseudocode ---
def _VOP3Op_V_MAD_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u16 = S0.u16 * S1.u16 + S2.u16
# --- compiled pseudocode ---
D0.u16 = S0.u16 * S1.u16 + S2.u16
# --- end pseudocode ---
def _VOP3Op_V_XAD_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = (S0.u32 ^ S1.u32) + S2.u32
# --- compiled pseudocode ---
D0.u32 = (S0.u32 ^ S1.u32) + S2.u32
# --- end pseudocode ---
def _VOP3Op_V_LSHL_ADD_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = (S0.u32 << S1.u32[4 : 0].u32) + S2.u32
# --- compiled pseudocode ---
D0.u32 = (S0.u32 << S1.u32[4 : 0].u32) + S2.u32
# --- end pseudocode ---
def _VOP3Op_V_ADD_LSHL_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = ((S0.u32 + S1.u32) << S2.u32[4 : 0].u32)
# --- compiled pseudocode ---
D0.u32 = ((S0.u32 + S1.u32) << S2.u32[4 : 0].u32)
# --- end pseudocode ---
def _VOP3Op_V_FMA_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = fma(S0.f16, S1.f16, S2.f16)
# --- compiled pseudocode ---
D0.f16 = fma(S0.f16, S1.f16, S2.f16)
# --- end pseudocode ---
def _VOP3Op_V_MIN3_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = v_min_f16(v_min_f16(S0.f16, S1.f16), S2.f16)
# --- compiled pseudocode ---
D0.f16 = v_min_f16(v_min_f16(S0.f16, S1.f16), S2.f16)
# --- end pseudocode ---
def _VOP3Op_V_MIN3_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i16 = v_min_i16(v_min_i16(S0.i16, S1.i16), S2.i16)
# --- compiled pseudocode ---
D0.i16 = v_min_i16(v_min_i16(S0.i16, S1.i16), S2.i16)
# --- end pseudocode ---
def _VOP3Op_V_MIN3_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u16 = v_min_u16(v_min_u16(S0.u16, S1.u16), S2.u16)
# --- compiled pseudocode ---
D0.u16 = v_min_u16(v_min_u16(S0.u16, S1.u16), S2.u16)
# --- end pseudocode ---
def _VOP3Op_V_MAX3_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = v_max_f16(v_max_f16(S0.f16, S1.f16), S2.f16)
# --- compiled pseudocode ---
D0.f16 = v_max_f16(v_max_f16(S0.f16, S1.f16), S2.f16)
# --- end pseudocode ---
def _VOP3Op_V_MAX3_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i16 = v_max_i16(v_max_i16(S0.i16, S1.i16), S2.i16)
# --- compiled pseudocode ---
D0.i16 = v_max_i16(v_max_i16(S0.i16, S1.i16), S2.i16)
# --- end pseudocode ---
def _VOP3Op_V_MAX3_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u16 = v_max_u16(v_max_u16(S0.u16, S1.u16), S2.u16)
# --- compiled pseudocode ---
D0.u16 = v_max_u16(v_max_u16(S0.u16, S1.u16), S2.u16)
# --- end pseudocode ---
def _VOP3Op_V_MED3_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if (isNAN(64'F(S0.f16)) || isNAN(64'F(S1.f16)) || isNAN(64'F(S2.f16))) then
# D0.f16 = v_min3_f16(S0.f16, S1.f16, S2.f16)
# elsif v_max3_f16(S0.f16, S1.f16, S2.f16) == S0.f16 then
# D0.f16 = v_max_f16(S1.f16, S2.f16)
# elsif v_max3_f16(S0.f16, S1.f16, S2.f16) == S1.f16 then
# D0.f16 = v_max_f16(S0.f16, S2.f16)
# else
# D0.f16 = v_max_f16(S0.f16, S1.f16)
# endif
# --- compiled pseudocode ---
if (isNAN(F(S0.f16)) or isNAN(F(S1.f16)) or isNAN(F(S2.f16))):
D0.f16 = v_min3_f16(S0.f16, S1.f16, S2.f16)
elif v_max3_f16(S0.f16, S1.f16, S2.f16) == S0.f16:
D0.f16 = v_max_f16(S1.f16, S2.f16)
elif v_max3_f16(S0.f16, S1.f16, S2.f16) == S1.f16:
D0.f16 = v_max_f16(S0.f16, S2.f16)
else:
D0.f16 = v_max_f16(S0.f16, S1.f16)
# --- end pseudocode ---
def _VOP3Op_V_MED3_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if v_max3_i16(S0.i16, S1.i16, S2.i16) == S0.i16 then
# D0.i16 = v_max_i16(S1.i16, S2.i16)
# elsif v_max3_i16(S0.i16, S1.i16, S2.i16) == S1.i16 then
# D0.i16 = v_max_i16(S0.i16, S2.i16)
# else
# D0.i16 = v_max_i16(S0.i16, S1.i16)
# endif
# --- compiled pseudocode ---
if v_max3_i16(S0.i16, S1.i16, S2.i16) == S0.i16:
D0.i16 = v_max_i16(S1.i16, S2.i16)
elif v_max3_i16(S0.i16, S1.i16, S2.i16) == S1.i16:
D0.i16 = v_max_i16(S0.i16, S2.i16)
else:
D0.i16 = v_max_i16(S0.i16, S1.i16)
# --- end pseudocode ---
def _VOP3Op_V_MED3_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# if v_max3_u16(S0.u16, S1.u16, S2.u16) == S0.u16 then
# D0.u16 = v_max_u16(S1.u16, S2.u16)
# elsif v_max3_u16(S0.u16, S1.u16, S2.u16) == S1.u16 then
# D0.u16 = v_max_u16(S0.u16, S2.u16)
# else
# D0.u16 = v_max_u16(S0.u16, S1.u16)
# endif
# --- compiled pseudocode ---
if v_max3_u16(S0.u16, S1.u16, S2.u16) == S0.u16:
D0.u16 = v_max_u16(S1.u16, S2.u16)
elif v_max3_u16(S0.u16, S1.u16, S2.u16) == S1.u16:
D0.u16 = v_max_u16(S0.u16, S2.u16)
else:
D0.u16 = v_max_u16(S0.u16, S1.u16)
# --- end pseudocode ---
def _VOP3Op_V_MAD_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i16 = S0.i16 * S1.i16 + S2.i16
# --- compiled pseudocode ---
D0.i16 = S0.i16 * S1.i16 + S2.i16
# --- end pseudocode ---
def _VOP3Op_V_DIV_FIXUP_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# sign_out = (sign(S1.f16) ^ sign(S2.f16));
# if isNAN(64'F(S2.f16)) then
# D0.f16 = 16'F(cvtToQuietNAN(64'F(S2.f16)))
# elsif isNAN(64'F(S1.f16)) then
# D0.f16 = 16'F(cvtToQuietNAN(64'F(S1.f16)))
# elsif ((64'F(S1.f16) == 0.0) && (64'F(S2.f16) == 0.0)) then
# // 0/0
# D0.f16 = 16'F(0xfe00)
# elsif ((64'F(abs(S1.f16)) == +INF) && (64'F(abs(S2.f16)) == +INF)) then
# // inf/inf
# D0.f16 = 16'F(0xfe00)
# elsif ((64'F(S1.f16) == 0.0) || (64'F(abs(S2.f16)) == +INF)) then
# // x/0, or inf/y
# D0.f16 = sign_out ? -INF.f16 : +INF.f16
# elsif ((64'F(abs(S1.f16)) == +INF) || (64'F(S2.f16) == 0.0)) then
# // x/inf, 0/y
# D0.f16 = sign_out ? -16'0.0 : 16'0.0
# else
# D0.f16 = sign_out ? -abs(S0.f16) : abs(S0.f16)
# endif
# --- compiled pseudocode ---
sign_out = (sign(S1.f16) ^ sign(S2.f16))
if isNAN(F(S2.f16)):
D0.f16 = F(cvtToQuietNAN(F(S2.f16)))
elif isNAN(F(S1.f16)):
D0.f16 = F(cvtToQuietNAN(F(S1.f16)))
elif ((F(S1.f16) == 0.0) and (F(S2.f16) == 0.0)):
D0.f16 = F(0xfe00)
elif ((F(abs(S1.f16)) == INF) and (F(abs(S2.f16)) == INF)):
D0.f16 = F(0xfe00)
elif ((F(S1.f16) == 0.0) or (F(abs(S2.f16)) == INF)):
D0.f16 = (((-INF).f16) if (sign_out) else (INF.f16))
elif ((F(abs(S1.f16)) == INF) or (F(S2.f16) == 0.0)):
D0.f16 = ((-0.0) if (sign_out) else (0.0))
else:
D0.f16 = ((-abs(S0.f16)) if (sign_out) else (abs(S0.f16)))
# --- end pseudocode ---
def _VOP3Op_V_ADD3_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = S0.u32 + S1.u32 + S2.u32
# --- compiled pseudocode ---
D0.u32 = S0.u32 + S1.u32 + S2.u32
# --- end pseudocode ---
def _VOP3Op_V_LSHL_OR_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = ((S0.u32 << S1.u32[4 : 0].u32) | S2.u32)
# --- compiled pseudocode ---
D0.u32 = ((S0.u32 << S1.u32[4 : 0].u32) | S2.u32)
# --- end pseudocode ---
def _VOP3Op_V_AND_OR_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = ((S0.u32 & S1.u32) | S2.u32)
# --- compiled pseudocode ---
D0.u32 = ((S0.u32 & S1.u32) | S2.u32)
# --- end pseudocode ---
def _VOP3Op_V_OR3_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = (S0.u32 | S1.u32 | S2.u32)
# --- compiled pseudocode ---
D0.u32 = (S0.u32 | S1.u32 | S2.u32)
# --- end pseudocode ---
def _VOP3Op_V_MAD_U32_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = 32'U(S0.u16) * 32'U(S1.u16) + S2.u32
# --- compiled pseudocode ---
D0.u32 = (S0.u16) * (S1.u16) + S2.u32
# --- end pseudocode ---
def _VOP3Op_V_MAD_I32_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = 32'I(S0.i16) * 32'I(S1.i16) + S2.i32
# --- compiled pseudocode ---
D0.i32 = (S0.i16) * (S1.i16) + S2.i32
# --- end pseudocode ---
def _VOP3Op_V_CNDMASK_B16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u16 = VCC.u64[laneId] ? S1.u16 : S0.u16
# --- compiled pseudocode ---
D0.u16 = ((S1.u16) if (VCC.u64[laneId]) else (S0.u16))
# --- end pseudocode ---
def _VOP3Op_V_MAXMIN_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = v_min_f32(v_max_f32(S0.f32, S1.f32), S2.f32)
# --- compiled pseudocode ---
D0.f32 = v_min_f32(v_max_f32(S0.f32, S1.f32), S2.f32)
# --- end pseudocode ---
def _VOP3Op_V_MINMAX_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = v_max_f32(v_min_f32(S0.f32, S1.f32), S2.f32)
# --- compiled pseudocode ---
D0.f32 = v_max_f32(v_min_f32(S0.f32, S1.f32), S2.f32)
# --- end pseudocode ---
def _VOP3Op_V_MAXMIN_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = v_min_f16(v_max_f16(S0.f16, S1.f16), S2.f16)
# --- compiled pseudocode ---
D0.f16 = v_min_f16(v_max_f16(S0.f16, S1.f16), S2.f16)
# --- end pseudocode ---
def _VOP3Op_V_MINMAX_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f16 = v_max_f16(v_min_f16(S0.f16, S1.f16), S2.f16)
# --- compiled pseudocode ---
D0.f16 = v_max_f16(v_min_f16(S0.f16, S1.f16), S2.f16)
# --- end pseudocode ---
def _VOP3Op_V_MAXMIN_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = v_min_u32(v_max_u32(S0.u32, S1.u32), S2.u32)
# --- compiled pseudocode ---
D0.u32 = v_min_u32(v_max_u32(S0.u32, S1.u32), S2.u32)
# --- end pseudocode ---
def _VOP3Op_V_MINMAX_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = v_max_u32(v_min_u32(S0.u32, S1.u32), S2.u32)
# --- compiled pseudocode ---
D0.u32 = v_max_u32(v_min_u32(S0.u32, S1.u32), S2.u32)
# --- end pseudocode ---
def _VOP3Op_V_MAXMIN_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = v_min_i32(v_max_i32(S0.i32, S1.i32), S2.i32)
# --- compiled pseudocode ---
D0.i32 = v_min_i32(v_max_i32(S0.i32, S1.i32), S2.i32)
# --- end pseudocode ---
def _VOP3Op_V_MINMAX_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = v_max_i32(v_min_i32(S0.i32, S1.i32), S2.i32)
# --- compiled pseudocode ---
D0.i32 = v_max_i32(v_min_i32(S0.i32, S1.i32), S2.i32)
# --- end pseudocode ---
def _VOP3Op_V_DOT2_F16_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = S2.f16;
# tmp += S0[15 : 0].f16 * S1[15 : 0].f16;
# tmp += S0[31 : 16].f16 * S1[31 : 16].f16;
# D0.f16 = tmp
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(S2.f16)
tmp += S0[15 : 0].f16 * S1[15 : 0].f16
tmp += S0[31 : 16].f16 * S1[31 : 16].f16
D0.f16 = tmp
# --- end pseudocode ---
def _VOP3Op_V_DOT2_BF16_BF16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = S2.bf16;
# tmp += S0[15 : 0].bf16 * S1[15 : 0].bf16;
# tmp += S0[31 : 16].bf16 * S1[31 : 16].bf16;
# D0.bf16 = tmp
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(S2.bf16)
tmp += S0[15 : 0].bf16 * S1[15 : 0].bf16
tmp += S0[31 : 16].bf16 * S1[31 : 16].bf16
D0.bf16 = tmp
# --- end pseudocode ---
def _VOP3Op_V_ADD_NC_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u16 = S0.u16 + S1.u16
# --- compiled pseudocode ---
D0.u16 = S0.u16 + S1.u16
# --- end pseudocode ---
def _VOP3Op_V_SUB_NC_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u16 = S0.u16 - S1.u16
# --- compiled pseudocode ---
D0.u16 = S0.u16 - S1.u16
# --- end pseudocode ---
def _VOP3Op_V_MUL_LO_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u16 = S0.u16 * S1.u16
# --- compiled pseudocode ---
D0.u16 = S0.u16 * S1.u16
# --- end pseudocode ---
def _VOP3Op_V_CVT_PK_I16_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# declare tmp : 32'B;
# tmp[31 : 16] = 16'B(v_cvt_i16_f32(S1.f32));
# tmp[15 : 0] = 16'B(v_cvt_i16_f32(S0.f32));
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[31 : 16] = (v_cvt_i16_f32(S1.f32))
tmp[15 : 0] = (v_cvt_i16_f32(S0.f32))
# --- end pseudocode ---
def _VOP3Op_V_CVT_PK_U16_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# declare tmp : 32'B;
# tmp[31 : 16] = 16'B(v_cvt_u16_f32(S1.f32));
# tmp[15 : 0] = 16'B(v_cvt_u16_f32(S0.f32));
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[31 : 16] = (v_cvt_u16_f32(S1.f32))
tmp[15 : 0] = (v_cvt_u16_f32(S0.f32))
# --- end pseudocode ---
def _VOP3Op_V_MAX_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u16 = S0.u16 >= S1.u16 ? S0.u16 : S1.u16
# --- compiled pseudocode ---
D0.u16 = ((S0.u16) if (S0.u16 >= S1.u16) else (S1.u16))
# --- end pseudocode ---
def _VOP3Op_V_MAX_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i16 = S0.i16 >= S1.i16 ? S0.i16 : S1.i16
# --- compiled pseudocode ---
D0.i16 = ((S0.i16) if (S0.i16 >= S1.i16) else (S1.i16))
# --- end pseudocode ---
def _VOP3Op_V_MIN_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u16 = S0.u16 < S1.u16 ? S0.u16 : S1.u16
# --- compiled pseudocode ---
D0.u16 = ((S0.u16) if (S0.u16 < S1.u16) else (S1.u16))
# --- end pseudocode ---
def _VOP3Op_V_MIN_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i16 = S0.i16 < S1.i16 ? S0.i16 : S1.i16
# --- compiled pseudocode ---
D0.i16 = ((S0.i16) if (S0.i16 < S1.i16) else (S1.i16))
# --- end pseudocode ---
def _VOP3Op_V_ADD_NC_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i16 = S0.i16 + S1.i16
# --- compiled pseudocode ---
D0.i16 = S0.i16 + S1.i16
# --- end pseudocode ---
def _VOP3Op_V_SUB_NC_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i16 = S0.i16 - S1.i16
# --- compiled pseudocode ---
D0.i16 = S0.i16 - S1.i16
# --- end pseudocode ---
def _VOP3Op_V_PACK_B32_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0[31 : 16].f16 = S1.f16;
# D0[15 : 0].f16 = S0.f16
# --- compiled pseudocode ---
D0[31 : 16].f16 = S1.f16
D0[15 : 0].f16 = S0.f16
# --- end pseudocode ---
def _VOP3Op_V_CVT_PK_NORM_I16_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# declare tmp : 32'B;
# tmp[15 : 0].i16 = f16_to_snorm(S0.f16);
# tmp[31 : 16].i16 = f16_to_snorm(S1.f16);
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[15 : 0].i16 = f16_to_snorm(S0.f16)
tmp[31 : 16].i16 = f16_to_snorm(S1.f16)
# --- end pseudocode ---
def _VOP3Op_V_CVT_PK_NORM_U16_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# declare tmp : 32'B;
# tmp[15 : 0].u16 = f16_to_unorm(S0.f16);
# tmp[31 : 16].u16 = f16_to_unorm(S1.f16);
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[15 : 0].u16 = f16_to_unorm(S0.f16)
tmp[31 : 16].u16 = f16_to_unorm(S1.f16)
# --- end pseudocode ---
def _VOP3Op_V_LDEXP_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f32 = S0.f32 * 2.0F ** S1.i32
# --- compiled pseudocode ---
D0.f32 = S0.f32 * 2.0 ** S1.i32
# --- end pseudocode ---
def _VOP3Op_V_BFM_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = (((1U << S0[4 : 0].u32) - 1U) << S1[4 : 0].u32)
# --- compiled pseudocode ---
D0.u32 = (((1 << S0[4 : 0].u32) - 1) << S1[4 : 0].u32)
# --- end pseudocode ---
def _VOP3Op_V_BCNT_U32_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = S1.u32;
# for i in 0 : 31 do
# tmp += S0[i].u32;
# // count i'th bit
# endfor;
# D0.u32 = tmp
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(S1.u32)
for i in range(0, int(31)+1):
tmp += S0[i].u32
D0.u32 = tmp
# --- end pseudocode ---
def _VOP3Op_V_CVT_PK_NORM_I16_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# declare tmp : 32'B;
# tmp[15 : 0].i16 = f32_to_snorm(S0.f32);
# tmp[31 : 16].i16 = f32_to_snorm(S1.f32);
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[15 : 0].i16 = f32_to_snorm(S0.f32)
tmp[31 : 16].i16 = f32_to_snorm(S1.f32)
# --- end pseudocode ---
def _VOP3Op_V_CVT_PK_NORM_U16_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# declare tmp : 32'B;
# tmp[15 : 0].u16 = f32_to_unorm(S0.f32);
# tmp[31 : 16].u16 = f32_to_unorm(S1.f32);
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[15 : 0].u16 = f32_to_unorm(S0.f32)
tmp[31 : 16].u16 = f32_to_unorm(S1.f32)
# --- end pseudocode ---
def _VOP3Op_V_CVT_PK_U16_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# declare tmp : 32'B;
# tmp[15 : 0].u16 = u32_to_u16(S0.u32);
# tmp[31 : 16].u16 = u32_to_u16(S1.u32);
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[15 : 0].u16 = u32_to_u16(S0.u32)
tmp[31 : 16].u16 = u32_to_u16(S1.u32)
# --- end pseudocode ---
def _VOP3Op_V_CVT_PK_I16_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# declare tmp : 32'B;
# tmp[15 : 0].i16 = i32_to_i16(S0.i32);
# tmp[31 : 16].i16 = i32_to_i16(S1.i32);
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[15 : 0].i16 = i32_to_i16(S0.i32)
tmp[31 : 16].i16 = i32_to_i16(S1.i32)
# --- end pseudocode ---
def _VOP3Op_V_SUB_NC_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = S0.i32 - S1.i32
# --- compiled pseudocode ---
D0.i32 = S0.i32 - S1.i32
# --- end pseudocode ---
def _VOP3Op_V_ADD_NC_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = S0.i32 + S1.i32
# --- compiled pseudocode ---
D0.i32 = S0.i32 + S1.i32
# --- end pseudocode ---
def _VOP3Op_V_ADD_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f64 = S0.f64 + S1.f64
# --- compiled pseudocode ---
D0.f64 = S0.f64 + S1.f64
# --- end pseudocode ---
def _VOP3Op_V_MUL_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f64 = S0.f64 * S1.f64
# --- compiled pseudocode ---
D0.f64 = S0.f64 * S1.f64
# --- end pseudocode ---
def _VOP3Op_V_MIN_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# // Version of comparison where -0.0 < +0.0, differs from IEEE
# if WAVE_MODE.IEEE then
# if isSignalNAN(S0.f64) then
# D0.f64 = cvtToQuietNAN(S0.f64)
# elsif isSignalNAN(S1.f64) then
# D0.f64 = cvtToQuietNAN(S1.f64)
# elsif isQuietNAN(S1.f64) then
# D0.f64 = S0.f64
# elsif isQuietNAN(S0.f64) then
# D0.f64 = S1.f64
# elsif LT_NEG_ZERO(S0.f64, S1.f64) then
# // NOTE: -0<+0 is TRUE in this comparison
# D0.f64 = S0.f64
# else
# D0.f64 = S1.f64
# endif
# else
# if isNAN(S1.f64) then
# D0.f64 = S0.f64
# elsif isNAN(S0.f64) then
# D0.f64 = S1.f64
# elsif LT_NEG_ZERO(S0.f64, S1.f64) then
# // NOTE: -0<+0 is TRUE in this comparison
# D0.f64 = S0.f64
# else
# D0.f64 = S1.f64
# endif
# endif;
# // Inequalities in the above pseudocode behave differently from IEEE
# --- compiled pseudocode ---
if WAVE_MODE.IEEE:
if isSignalNAN(S0.f64):
D0.f64 = cvtToQuietNAN(S0.f64)
elif isSignalNAN(S1.f64):
D0.f64 = cvtToQuietNAN(S1.f64)
elif isQuietNAN(S1.f64):
D0.f64 = S0.f64
elif isQuietNAN(S0.f64):
D0.f64 = S1.f64
elif LT_NEG_ZERO(S0.f64, S1.f64):
D0.f64 = S0.f64
else:
D0.f64 = S1.f64
else:
if isNAN(S1.f64):
D0.f64 = S0.f64
elif isNAN(S0.f64):
D0.f64 = S1.f64
elif LT_NEG_ZERO(S0.f64, S1.f64):
D0.f64 = S0.f64
else:
D0.f64 = S1.f64
# --- end pseudocode ---
def _VOP3Op_V_MAX_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# // Version of comparison where +0.0 > -0.0, differs from IEEE
# if WAVE_MODE.IEEE then
# if isSignalNAN(S0.f64) then
# D0.f64 = cvtToQuietNAN(S0.f64)
# elsif isSignalNAN(S1.f64) then
# D0.f64 = cvtToQuietNAN(S1.f64)
# elsif isQuietNAN(S1.f64) then
# D0.f64 = S0.f64
# elsif isQuietNAN(S0.f64) then
# D0.f64 = S1.f64
# elsif GT_NEG_ZERO(S0.f64, S1.f64) then
# // NOTE: +0>-0 is TRUE in this comparison
# D0.f64 = S0.f64
# else
# D0.f64 = S1.f64
# endif
# else
# if isNAN(S1.f64) then
# D0.f64 = S0.f64
# elsif isNAN(S0.f64) then
# D0.f64 = S1.f64
# elsif GT_NEG_ZERO(S0.f64, S1.f64) then
# // NOTE: +0>-0 is TRUE in this comparison
# D0.f64 = S0.f64
# else
# D0.f64 = S1.f64
# endif
# endif;
# // Inequalities in the above pseudocode behave differently from IEEE
# --- compiled pseudocode ---
if WAVE_MODE.IEEE:
if isSignalNAN(S0.f64):
D0.f64 = cvtToQuietNAN(S0.f64)
elif isSignalNAN(S1.f64):
D0.f64 = cvtToQuietNAN(S1.f64)
elif isQuietNAN(S1.f64):
D0.f64 = S0.f64
elif isQuietNAN(S0.f64):
D0.f64 = S1.f64
elif GT_NEG_ZERO(S0.f64, S1.f64):
D0.f64 = S0.f64
else:
D0.f64 = S1.f64
else:
if isNAN(S1.f64):
D0.f64 = S0.f64
elif isNAN(S0.f64):
D0.f64 = S1.f64
elif GT_NEG_ZERO(S0.f64, S1.f64):
D0.f64 = S0.f64
else:
D0.f64 = S1.f64
# --- end pseudocode ---
def _VOP3Op_V_LDEXP_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.f64 = S0.f64 * 2.0 ** S1.i32
# --- compiled pseudocode ---
D0.f64 = S0.f64 * 2.0 ** S1.i32
# --- end pseudocode ---
def _VOP3Op_V_MUL_LO_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = S0.u32 * S1.u32
# --- compiled pseudocode ---
D0.u32 = S0.u32 * S1.u32
# --- end pseudocode ---
def _VOP3Op_V_MUL_HI_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u32 = 32'U((64'U(S0.u32) * 64'U(S1.u32)) >> 32U)
# --- compiled pseudocode ---
D0.u32 = (((S0.u32) * (S1.u32)) >> 32)
# --- end pseudocode ---
def _VOP3Op_V_MUL_HI_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i32 = 32'I((64'I(S0.i32) * 64'I(S1.i32)) >> 32U)
# --- compiled pseudocode ---
D0.i32 = (((S0.i32) * (S1.i32)) >> 32)
# --- end pseudocode ---
def _VOP3Op_V_LSHLREV_B16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u16 = (S1.u16 << S0[3 : 0].u32)
# --- compiled pseudocode ---
D0.u16 = (S1.u16 << S0[3 : 0].u32)
# --- end pseudocode ---
def _VOP3Op_V_LSHRREV_B16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u16 = (S1.u16 >> S0[3 : 0].u32)
# --- compiled pseudocode ---
D0.u16 = (S1.u16 >> S0[3 : 0].u32)
# --- end pseudocode ---
def _VOP3Op_V_ASHRREV_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i16 = (S1.i16 >> S0[3 : 0].u32)
# --- compiled pseudocode ---
D0.i16 = (S1.i16 >> S0[3 : 0].u32)
# --- end pseudocode ---
def _VOP3Op_V_LSHLREV_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64 = (S1.u64 << S0[5 : 0].u32)
# --- compiled pseudocode ---
D0.u64 = (S1.u64 << S0[5 : 0].u32)
# --- end pseudocode ---
def _VOP3Op_V_LSHRREV_B64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64 = (S1.u64 >> S0[5 : 0].u32)
# --- compiled pseudocode ---
D0.u64 = (S1.u64 >> S0[5 : 0].u32)
# --- end pseudocode ---
def _VOP3Op_V_ASHRREV_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.i64 = (S1.i64 >> S0[5 : 0].u32)
# --- compiled pseudocode ---
D0.i64 = (S1.i64 >> S0[5 : 0].u32)
# --- end pseudocode ---
def _VOP3Op_V_READLANE_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# declare lane : 32'U;
# if WAVE32 then
# lane = S1.u32[4 : 0].u32;
# // Lane select for wave32
# else
# lane = S1.u32[5 : 0].u32;
# // Lane select for wave64
# endif;
# D0.b32 = VGPR[lane][SRC0.u32]
# --- compiled pseudocode ---
if WAVE32:
lane = S1.u32[4 : 0].u32
else:
lane = S1.u32[5 : 0].u32
D0.b32 = VGPR[lane][SRC0.u32]
# --- end pseudocode ---
def _VOP3Op_V_AND_B16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u16 = (S0.u16 & S1.u16)
# --- compiled pseudocode ---
D0.u16 = (S0.u16 & S1.u16)
# --- end pseudocode ---
def _VOP3Op_V_OR_B16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u16 = (S0.u16 | S1.u16)
# --- compiled pseudocode ---
D0.u16 = (S0.u16 | S1.u16)
# --- end pseudocode ---
def _VOP3Op_V_XOR_B16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u16 = (S0.u16 ^ S1.u16)
# --- compiled pseudocode ---
D0.u16 = (S0.u16 ^ S1.u16)
# --- end pseudocode ---
VOP3Op_FUNCTIONS = {
VOP3Op.V_CMP_F_F16: _VOP3Op_V_CMP_F_F16,
VOP3Op.V_CMP_LT_F16: _VOP3Op_V_CMP_LT_F16,
VOP3Op.V_CMP_EQ_F16: _VOP3Op_V_CMP_EQ_F16,
VOP3Op.V_CMP_LE_F16: _VOP3Op_V_CMP_LE_F16,
VOP3Op.V_CMP_GT_F16: _VOP3Op_V_CMP_GT_F16,
VOP3Op.V_CMP_LG_F16: _VOP3Op_V_CMP_LG_F16,
VOP3Op.V_CMP_GE_F16: _VOP3Op_V_CMP_GE_F16,
VOP3Op.V_CMP_O_F16: _VOP3Op_V_CMP_O_F16,
VOP3Op.V_CMP_U_F16: _VOP3Op_V_CMP_U_F16,
VOP3Op.V_CMP_NGE_F16: _VOP3Op_V_CMP_NGE_F16,
VOP3Op.V_CMP_NLG_F16: _VOP3Op_V_CMP_NLG_F16,
VOP3Op.V_CMP_NGT_F16: _VOP3Op_V_CMP_NGT_F16,
VOP3Op.V_CMP_NLE_F16: _VOP3Op_V_CMP_NLE_F16,
VOP3Op.V_CMP_NEQ_F16: _VOP3Op_V_CMP_NEQ_F16,
VOP3Op.V_CMP_NLT_F16: _VOP3Op_V_CMP_NLT_F16,
VOP3Op.V_CMP_T_F16: _VOP3Op_V_CMP_T_F16,
VOP3Op.V_CMP_F_F32: _VOP3Op_V_CMP_F_F32,
VOP3Op.V_CMP_LT_F32: _VOP3Op_V_CMP_LT_F32,
VOP3Op.V_CMP_EQ_F32: _VOP3Op_V_CMP_EQ_F32,
VOP3Op.V_CMP_LE_F32: _VOP3Op_V_CMP_LE_F32,
VOP3Op.V_CMP_GT_F32: _VOP3Op_V_CMP_GT_F32,
VOP3Op.V_CMP_LG_F32: _VOP3Op_V_CMP_LG_F32,
VOP3Op.V_CMP_GE_F32: _VOP3Op_V_CMP_GE_F32,
VOP3Op.V_CMP_O_F32: _VOP3Op_V_CMP_O_F32,
VOP3Op.V_CMP_U_F32: _VOP3Op_V_CMP_U_F32,
VOP3Op.V_CMP_NGE_F32: _VOP3Op_V_CMP_NGE_F32,
VOP3Op.V_CMP_NLG_F32: _VOP3Op_V_CMP_NLG_F32,
VOP3Op.V_CMP_NGT_F32: _VOP3Op_V_CMP_NGT_F32,
VOP3Op.V_CMP_NLE_F32: _VOP3Op_V_CMP_NLE_F32,
VOP3Op.V_CMP_NEQ_F32: _VOP3Op_V_CMP_NEQ_F32,
VOP3Op.V_CMP_NLT_F32: _VOP3Op_V_CMP_NLT_F32,
VOP3Op.V_CMP_T_F32: _VOP3Op_V_CMP_T_F32,
VOP3Op.V_CMP_F_F64: _VOP3Op_V_CMP_F_F64,
VOP3Op.V_CMP_LT_F64: _VOP3Op_V_CMP_LT_F64,
VOP3Op.V_CMP_EQ_F64: _VOP3Op_V_CMP_EQ_F64,
VOP3Op.V_CMP_LE_F64: _VOP3Op_V_CMP_LE_F64,
VOP3Op.V_CMP_GT_F64: _VOP3Op_V_CMP_GT_F64,
VOP3Op.V_CMP_LG_F64: _VOP3Op_V_CMP_LG_F64,
VOP3Op.V_CMP_GE_F64: _VOP3Op_V_CMP_GE_F64,
VOP3Op.V_CMP_O_F64: _VOP3Op_V_CMP_O_F64,
VOP3Op.V_CMP_U_F64: _VOP3Op_V_CMP_U_F64,
VOP3Op.V_CMP_NGE_F64: _VOP3Op_V_CMP_NGE_F64,
VOP3Op.V_CMP_NLG_F64: _VOP3Op_V_CMP_NLG_F64,
VOP3Op.V_CMP_NGT_F64: _VOP3Op_V_CMP_NGT_F64,
VOP3Op.V_CMP_NLE_F64: _VOP3Op_V_CMP_NLE_F64,
VOP3Op.V_CMP_NEQ_F64: _VOP3Op_V_CMP_NEQ_F64,
VOP3Op.V_CMP_NLT_F64: _VOP3Op_V_CMP_NLT_F64,
VOP3Op.V_CMP_T_F64: _VOP3Op_V_CMP_T_F64,
VOP3Op.V_CMP_LT_I16: _VOP3Op_V_CMP_LT_I16,
VOP3Op.V_CMP_EQ_I16: _VOP3Op_V_CMP_EQ_I16,
VOP3Op.V_CMP_LE_I16: _VOP3Op_V_CMP_LE_I16,
VOP3Op.V_CMP_GT_I16: _VOP3Op_V_CMP_GT_I16,
VOP3Op.V_CMP_NE_I16: _VOP3Op_V_CMP_NE_I16,
VOP3Op.V_CMP_GE_I16: _VOP3Op_V_CMP_GE_I16,
VOP3Op.V_CMP_LT_U16: _VOP3Op_V_CMP_LT_U16,
VOP3Op.V_CMP_EQ_U16: _VOP3Op_V_CMP_EQ_U16,
VOP3Op.V_CMP_LE_U16: _VOP3Op_V_CMP_LE_U16,
VOP3Op.V_CMP_GT_U16: _VOP3Op_V_CMP_GT_U16,
VOP3Op.V_CMP_NE_U16: _VOP3Op_V_CMP_NE_U16,
VOP3Op.V_CMP_GE_U16: _VOP3Op_V_CMP_GE_U16,
VOP3Op.V_CMP_F_I32: _VOP3Op_V_CMP_F_I32,
VOP3Op.V_CMP_LT_I32: _VOP3Op_V_CMP_LT_I32,
VOP3Op.V_CMP_EQ_I32: _VOP3Op_V_CMP_EQ_I32,
VOP3Op.V_CMP_LE_I32: _VOP3Op_V_CMP_LE_I32,
VOP3Op.V_CMP_GT_I32: _VOP3Op_V_CMP_GT_I32,
VOP3Op.V_CMP_NE_I32: _VOP3Op_V_CMP_NE_I32,
VOP3Op.V_CMP_GE_I32: _VOP3Op_V_CMP_GE_I32,
VOP3Op.V_CMP_T_I32: _VOP3Op_V_CMP_T_I32,
VOP3Op.V_CMP_F_U32: _VOP3Op_V_CMP_F_U32,
VOP3Op.V_CMP_LT_U32: _VOP3Op_V_CMP_LT_U32,
VOP3Op.V_CMP_EQ_U32: _VOP3Op_V_CMP_EQ_U32,
VOP3Op.V_CMP_LE_U32: _VOP3Op_V_CMP_LE_U32,
VOP3Op.V_CMP_GT_U32: _VOP3Op_V_CMP_GT_U32,
VOP3Op.V_CMP_NE_U32: _VOP3Op_V_CMP_NE_U32,
VOP3Op.V_CMP_GE_U32: _VOP3Op_V_CMP_GE_U32,
VOP3Op.V_CMP_T_U32: _VOP3Op_V_CMP_T_U32,
VOP3Op.V_CMP_F_I64: _VOP3Op_V_CMP_F_I64,
VOP3Op.V_CMP_LT_I64: _VOP3Op_V_CMP_LT_I64,
VOP3Op.V_CMP_EQ_I64: _VOP3Op_V_CMP_EQ_I64,
VOP3Op.V_CMP_LE_I64: _VOP3Op_V_CMP_LE_I64,
VOP3Op.V_CMP_GT_I64: _VOP3Op_V_CMP_GT_I64,
VOP3Op.V_CMP_NE_I64: _VOP3Op_V_CMP_NE_I64,
VOP3Op.V_CMP_GE_I64: _VOP3Op_V_CMP_GE_I64,
VOP3Op.V_CMP_T_I64: _VOP3Op_V_CMP_T_I64,
VOP3Op.V_CMP_F_U64: _VOP3Op_V_CMP_F_U64,
VOP3Op.V_CMP_LT_U64: _VOP3Op_V_CMP_LT_U64,
VOP3Op.V_CMP_EQ_U64: _VOP3Op_V_CMP_EQ_U64,
VOP3Op.V_CMP_LE_U64: _VOP3Op_V_CMP_LE_U64,
VOP3Op.V_CMP_GT_U64: _VOP3Op_V_CMP_GT_U64,
VOP3Op.V_CMP_NE_U64: _VOP3Op_V_CMP_NE_U64,
VOP3Op.V_CMP_GE_U64: _VOP3Op_V_CMP_GE_U64,
VOP3Op.V_CMP_T_U64: _VOP3Op_V_CMP_T_U64,
VOP3Op.V_CMP_CLASS_F16: _VOP3Op_V_CMP_CLASS_F16,
VOP3Op.V_CMP_CLASS_F32: _VOP3Op_V_CMP_CLASS_F32,
VOP3Op.V_CMP_CLASS_F64: _VOP3Op_V_CMP_CLASS_F64,
VOP3Op.V_CMPX_F_F16: _VOP3Op_V_CMPX_F_F16,
VOP3Op.V_CMPX_LT_F16: _VOP3Op_V_CMPX_LT_F16,
VOP3Op.V_CMPX_EQ_F16: _VOP3Op_V_CMPX_EQ_F16,
VOP3Op.V_CMPX_LE_F16: _VOP3Op_V_CMPX_LE_F16,
VOP3Op.V_CMPX_GT_F16: _VOP3Op_V_CMPX_GT_F16,
VOP3Op.V_CMPX_LG_F16: _VOP3Op_V_CMPX_LG_F16,
VOP3Op.V_CMPX_GE_F16: _VOP3Op_V_CMPX_GE_F16,
VOP3Op.V_CMPX_O_F16: _VOP3Op_V_CMPX_O_F16,
VOP3Op.V_CMPX_U_F16: _VOP3Op_V_CMPX_U_F16,
VOP3Op.V_CMPX_NGE_F16: _VOP3Op_V_CMPX_NGE_F16,
VOP3Op.V_CMPX_NLG_F16: _VOP3Op_V_CMPX_NLG_F16,
VOP3Op.V_CMPX_NGT_F16: _VOP3Op_V_CMPX_NGT_F16,
VOP3Op.V_CMPX_NLE_F16: _VOP3Op_V_CMPX_NLE_F16,
VOP3Op.V_CMPX_NEQ_F16: _VOP3Op_V_CMPX_NEQ_F16,
VOP3Op.V_CMPX_NLT_F16: _VOP3Op_V_CMPX_NLT_F16,
VOP3Op.V_CMPX_T_F16: _VOP3Op_V_CMPX_T_F16,
VOP3Op.V_CMPX_F_F32: _VOP3Op_V_CMPX_F_F32,
VOP3Op.V_CMPX_LT_F32: _VOP3Op_V_CMPX_LT_F32,
VOP3Op.V_CMPX_EQ_F32: _VOP3Op_V_CMPX_EQ_F32,
VOP3Op.V_CMPX_LE_F32: _VOP3Op_V_CMPX_LE_F32,
VOP3Op.V_CMPX_GT_F32: _VOP3Op_V_CMPX_GT_F32,
VOP3Op.V_CMPX_LG_F32: _VOP3Op_V_CMPX_LG_F32,
VOP3Op.V_CMPX_GE_F32: _VOP3Op_V_CMPX_GE_F32,
VOP3Op.V_CMPX_O_F32: _VOP3Op_V_CMPX_O_F32,
VOP3Op.V_CMPX_U_F32: _VOP3Op_V_CMPX_U_F32,
VOP3Op.V_CMPX_NGE_F32: _VOP3Op_V_CMPX_NGE_F32,
VOP3Op.V_CMPX_NLG_F32: _VOP3Op_V_CMPX_NLG_F32,
VOP3Op.V_CMPX_NGT_F32: _VOP3Op_V_CMPX_NGT_F32,
VOP3Op.V_CMPX_NLE_F32: _VOP3Op_V_CMPX_NLE_F32,
VOP3Op.V_CMPX_NEQ_F32: _VOP3Op_V_CMPX_NEQ_F32,
VOP3Op.V_CMPX_NLT_F32: _VOP3Op_V_CMPX_NLT_F32,
VOP3Op.V_CMPX_T_F32: _VOP3Op_V_CMPX_T_F32,
VOP3Op.V_CMPX_F_F64: _VOP3Op_V_CMPX_F_F64,
VOP3Op.V_CMPX_LT_F64: _VOP3Op_V_CMPX_LT_F64,
VOP3Op.V_CMPX_EQ_F64: _VOP3Op_V_CMPX_EQ_F64,
VOP3Op.V_CMPX_LE_F64: _VOP3Op_V_CMPX_LE_F64,
VOP3Op.V_CMPX_GT_F64: _VOP3Op_V_CMPX_GT_F64,
VOP3Op.V_CMPX_LG_F64: _VOP3Op_V_CMPX_LG_F64,
VOP3Op.V_CMPX_GE_F64: _VOP3Op_V_CMPX_GE_F64,
VOP3Op.V_CMPX_O_F64: _VOP3Op_V_CMPX_O_F64,
VOP3Op.V_CMPX_U_F64: _VOP3Op_V_CMPX_U_F64,
VOP3Op.V_CMPX_NGE_F64: _VOP3Op_V_CMPX_NGE_F64,
VOP3Op.V_CMPX_NLG_F64: _VOP3Op_V_CMPX_NLG_F64,
VOP3Op.V_CMPX_NGT_F64: _VOP3Op_V_CMPX_NGT_F64,
VOP3Op.V_CMPX_NLE_F64: _VOP3Op_V_CMPX_NLE_F64,
VOP3Op.V_CMPX_NEQ_F64: _VOP3Op_V_CMPX_NEQ_F64,
VOP3Op.V_CMPX_NLT_F64: _VOP3Op_V_CMPX_NLT_F64,
VOP3Op.V_CMPX_T_F64: _VOP3Op_V_CMPX_T_F64,
VOP3Op.V_CMPX_LT_I16: _VOP3Op_V_CMPX_LT_I16,
VOP3Op.V_CMPX_EQ_I16: _VOP3Op_V_CMPX_EQ_I16,
VOP3Op.V_CMPX_LE_I16: _VOP3Op_V_CMPX_LE_I16,
VOP3Op.V_CMPX_GT_I16: _VOP3Op_V_CMPX_GT_I16,
VOP3Op.V_CMPX_NE_I16: _VOP3Op_V_CMPX_NE_I16,
VOP3Op.V_CMPX_GE_I16: _VOP3Op_V_CMPX_GE_I16,
VOP3Op.V_CMPX_LT_U16: _VOP3Op_V_CMPX_LT_U16,
VOP3Op.V_CMPX_EQ_U16: _VOP3Op_V_CMPX_EQ_U16,
VOP3Op.V_CMPX_LE_U16: _VOP3Op_V_CMPX_LE_U16,
VOP3Op.V_CMPX_GT_U16: _VOP3Op_V_CMPX_GT_U16,
VOP3Op.V_CMPX_NE_U16: _VOP3Op_V_CMPX_NE_U16,
VOP3Op.V_CMPX_GE_U16: _VOP3Op_V_CMPX_GE_U16,
VOP3Op.V_CMPX_F_I32: _VOP3Op_V_CMPX_F_I32,
VOP3Op.V_CMPX_LT_I32: _VOP3Op_V_CMPX_LT_I32,
VOP3Op.V_CMPX_EQ_I32: _VOP3Op_V_CMPX_EQ_I32,
VOP3Op.V_CMPX_LE_I32: _VOP3Op_V_CMPX_LE_I32,
VOP3Op.V_CMPX_GT_I32: _VOP3Op_V_CMPX_GT_I32,
VOP3Op.V_CMPX_NE_I32: _VOP3Op_V_CMPX_NE_I32,
VOP3Op.V_CMPX_GE_I32: _VOP3Op_V_CMPX_GE_I32,
VOP3Op.V_CMPX_T_I32: _VOP3Op_V_CMPX_T_I32,
VOP3Op.V_CMPX_F_U32: _VOP3Op_V_CMPX_F_U32,
VOP3Op.V_CMPX_LT_U32: _VOP3Op_V_CMPX_LT_U32,
VOP3Op.V_CMPX_EQ_U32: _VOP3Op_V_CMPX_EQ_U32,
VOP3Op.V_CMPX_LE_U32: _VOP3Op_V_CMPX_LE_U32,
VOP3Op.V_CMPX_GT_U32: _VOP3Op_V_CMPX_GT_U32,
VOP3Op.V_CMPX_NE_U32: _VOP3Op_V_CMPX_NE_U32,
VOP3Op.V_CMPX_GE_U32: _VOP3Op_V_CMPX_GE_U32,
VOP3Op.V_CMPX_T_U32: _VOP3Op_V_CMPX_T_U32,
VOP3Op.V_CMPX_F_I64: _VOP3Op_V_CMPX_F_I64,
VOP3Op.V_CMPX_LT_I64: _VOP3Op_V_CMPX_LT_I64,
VOP3Op.V_CMPX_EQ_I64: _VOP3Op_V_CMPX_EQ_I64,
VOP3Op.V_CMPX_LE_I64: _VOP3Op_V_CMPX_LE_I64,
VOP3Op.V_CMPX_GT_I64: _VOP3Op_V_CMPX_GT_I64,
VOP3Op.V_CMPX_NE_I64: _VOP3Op_V_CMPX_NE_I64,
VOP3Op.V_CMPX_GE_I64: _VOP3Op_V_CMPX_GE_I64,
VOP3Op.V_CMPX_T_I64: _VOP3Op_V_CMPX_T_I64,
VOP3Op.V_CMPX_F_U64: _VOP3Op_V_CMPX_F_U64,
VOP3Op.V_CMPX_LT_U64: _VOP3Op_V_CMPX_LT_U64,
VOP3Op.V_CMPX_EQ_U64: _VOP3Op_V_CMPX_EQ_U64,
VOP3Op.V_CMPX_LE_U64: _VOP3Op_V_CMPX_LE_U64,
VOP3Op.V_CMPX_GT_U64: _VOP3Op_V_CMPX_GT_U64,
VOP3Op.V_CMPX_NE_U64: _VOP3Op_V_CMPX_NE_U64,
VOP3Op.V_CMPX_GE_U64: _VOP3Op_V_CMPX_GE_U64,
VOP3Op.V_CMPX_T_U64: _VOP3Op_V_CMPX_T_U64,
VOP3Op.V_CMPX_CLASS_F16: _VOP3Op_V_CMPX_CLASS_F16,
VOP3Op.V_CMPX_CLASS_F32: _VOP3Op_V_CMPX_CLASS_F32,
VOP3Op.V_CMPX_CLASS_F64: _VOP3Op_V_CMPX_CLASS_F64,
VOP3Op.V_MOV_B32: _VOP3Op_V_MOV_B32,
VOP3Op.V_READFIRSTLANE_B32: _VOP3Op_V_READFIRSTLANE_B32,
VOP3Op.V_CVT_I32_F64: _VOP3Op_V_CVT_I32_F64,
VOP3Op.V_CVT_F64_I32: _VOP3Op_V_CVT_F64_I32,
VOP3Op.V_CVT_F32_I32: _VOP3Op_V_CVT_F32_I32,
VOP3Op.V_CVT_F32_U32: _VOP3Op_V_CVT_F32_U32,
VOP3Op.V_CVT_U32_F32: _VOP3Op_V_CVT_U32_F32,
VOP3Op.V_CVT_I32_F32: _VOP3Op_V_CVT_I32_F32,
VOP3Op.V_CVT_F16_F32: _VOP3Op_V_CVT_F16_F32,
VOP3Op.V_CVT_F32_F16: _VOP3Op_V_CVT_F32_F16,
VOP3Op.V_CVT_NEAREST_I32_F32: _VOP3Op_V_CVT_NEAREST_I32_F32,
VOP3Op.V_CVT_FLOOR_I32_F32: _VOP3Op_V_CVT_FLOOR_I32_F32,
VOP3Op.V_CVT_F32_F64: _VOP3Op_V_CVT_F32_F64,
VOP3Op.V_CVT_F64_F32: _VOP3Op_V_CVT_F64_F32,
VOP3Op.V_CVT_F32_UBYTE0: _VOP3Op_V_CVT_F32_UBYTE0,
VOP3Op.V_CVT_F32_UBYTE1: _VOP3Op_V_CVT_F32_UBYTE1,
VOP3Op.V_CVT_F32_UBYTE2: _VOP3Op_V_CVT_F32_UBYTE2,
VOP3Op.V_CVT_F32_UBYTE3: _VOP3Op_V_CVT_F32_UBYTE3,
VOP3Op.V_CVT_U32_F64: _VOP3Op_V_CVT_U32_F64,
VOP3Op.V_CVT_F64_U32: _VOP3Op_V_CVT_F64_U32,
VOP3Op.V_TRUNC_F64: _VOP3Op_V_TRUNC_F64,
VOP3Op.V_CEIL_F64: _VOP3Op_V_CEIL_F64,
VOP3Op.V_RNDNE_F64: _VOP3Op_V_RNDNE_F64,
VOP3Op.V_FLOOR_F64: _VOP3Op_V_FLOOR_F64,
VOP3Op.V_MOV_B16: _VOP3Op_V_MOV_B16,
VOP3Op.V_FRACT_F32: _VOP3Op_V_FRACT_F32,
VOP3Op.V_TRUNC_F32: _VOP3Op_V_TRUNC_F32,
VOP3Op.V_CEIL_F32: _VOP3Op_V_CEIL_F32,
VOP3Op.V_RNDNE_F32: _VOP3Op_V_RNDNE_F32,
VOP3Op.V_FLOOR_F32: _VOP3Op_V_FLOOR_F32,
VOP3Op.V_EXP_F32: _VOP3Op_V_EXP_F32,
VOP3Op.V_LOG_F32: _VOP3Op_V_LOG_F32,
VOP3Op.V_RCP_F32: _VOP3Op_V_RCP_F32,
VOP3Op.V_RCP_IFLAG_F32: _VOP3Op_V_RCP_IFLAG_F32,
VOP3Op.V_RSQ_F32: _VOP3Op_V_RSQ_F32,
VOP3Op.V_RCP_F64: _VOP3Op_V_RCP_F64,
VOP3Op.V_RSQ_F64: _VOP3Op_V_RSQ_F64,
VOP3Op.V_SQRT_F32: _VOP3Op_V_SQRT_F32,
VOP3Op.V_SQRT_F64: _VOP3Op_V_SQRT_F64,
VOP3Op.V_SIN_F32: _VOP3Op_V_SIN_F32,
VOP3Op.V_COS_F32: _VOP3Op_V_COS_F32,
VOP3Op.V_NOT_B32: _VOP3Op_V_NOT_B32,
VOP3Op.V_BFREV_B32: _VOP3Op_V_BFREV_B32,
VOP3Op.V_CLZ_I32_U32: _VOP3Op_V_CLZ_I32_U32,
VOP3Op.V_CTZ_I32_B32: _VOP3Op_V_CTZ_I32_B32,
VOP3Op.V_CLS_I32: _VOP3Op_V_CLS_I32,
VOP3Op.V_FREXP_EXP_I32_F64: _VOP3Op_V_FREXP_EXP_I32_F64,
VOP3Op.V_FREXP_MANT_F64: _VOP3Op_V_FREXP_MANT_F64,
VOP3Op.V_FRACT_F64: _VOP3Op_V_FRACT_F64,
VOP3Op.V_FREXP_EXP_I32_F32: _VOP3Op_V_FREXP_EXP_I32_F32,
VOP3Op.V_FREXP_MANT_F32: _VOP3Op_V_FREXP_MANT_F32,
VOP3Op.V_MOVRELS_B32: _VOP3Op_V_MOVRELS_B32,
VOP3Op.V_CVT_F16_U16: _VOP3Op_V_CVT_F16_U16,
VOP3Op.V_CVT_F16_I16: _VOP3Op_V_CVT_F16_I16,
VOP3Op.V_CVT_U16_F16: _VOP3Op_V_CVT_U16_F16,
VOP3Op.V_CVT_I16_F16: _VOP3Op_V_CVT_I16_F16,
VOP3Op.V_RCP_F16: _VOP3Op_V_RCP_F16,
VOP3Op.V_SQRT_F16: _VOP3Op_V_SQRT_F16,
VOP3Op.V_RSQ_F16: _VOP3Op_V_RSQ_F16,
VOP3Op.V_LOG_F16: _VOP3Op_V_LOG_F16,
VOP3Op.V_EXP_F16: _VOP3Op_V_EXP_F16,
VOP3Op.V_FREXP_MANT_F16: _VOP3Op_V_FREXP_MANT_F16,
VOP3Op.V_FREXP_EXP_I16_F16: _VOP3Op_V_FREXP_EXP_I16_F16,
VOP3Op.V_FLOOR_F16: _VOP3Op_V_FLOOR_F16,
VOP3Op.V_CEIL_F16: _VOP3Op_V_CEIL_F16,
VOP3Op.V_TRUNC_F16: _VOP3Op_V_TRUNC_F16,
VOP3Op.V_RNDNE_F16: _VOP3Op_V_RNDNE_F16,
VOP3Op.V_FRACT_F16: _VOP3Op_V_FRACT_F16,
VOP3Op.V_SIN_F16: _VOP3Op_V_SIN_F16,
VOP3Op.V_COS_F16: _VOP3Op_V_COS_F16,
VOP3Op.V_CVT_NORM_I16_F16: _VOP3Op_V_CVT_NORM_I16_F16,
VOP3Op.V_CVT_NORM_U16_F16: _VOP3Op_V_CVT_NORM_U16_F16,
VOP3Op.V_NOT_B16: _VOP3Op_V_NOT_B16,
VOP3Op.V_CVT_I32_I16: _VOP3Op_V_CVT_I32_I16,
VOP3Op.V_CVT_U32_U16: _VOP3Op_V_CVT_U32_U16,
VOP3Op.V_CNDMASK_B32: _VOP3Op_V_CNDMASK_B32,
VOP3Op.V_ADD_F32: _VOP3Op_V_ADD_F32,
VOP3Op.V_SUB_F32: _VOP3Op_V_SUB_F32,
VOP3Op.V_SUBREV_F32: _VOP3Op_V_SUBREV_F32,
VOP3Op.V_FMAC_DX9_ZERO_F32: _VOP3Op_V_FMAC_DX9_ZERO_F32,
VOP3Op.V_MUL_DX9_ZERO_F32: _VOP3Op_V_MUL_DX9_ZERO_F32,
VOP3Op.V_MUL_F32: _VOP3Op_V_MUL_F32,
VOP3Op.V_MUL_I32_I24: _VOP3Op_V_MUL_I32_I24,
VOP3Op.V_MUL_HI_I32_I24: _VOP3Op_V_MUL_HI_I32_I24,
VOP3Op.V_MUL_U32_U24: _VOP3Op_V_MUL_U32_U24,
VOP3Op.V_MUL_HI_U32_U24: _VOP3Op_V_MUL_HI_U32_U24,
VOP3Op.V_MIN_F32: _VOP3Op_V_MIN_F32,
VOP3Op.V_MAX_F32: _VOP3Op_V_MAX_F32,
VOP3Op.V_MIN_I32: _VOP3Op_V_MIN_I32,
VOP3Op.V_MAX_I32: _VOP3Op_V_MAX_I32,
VOP3Op.V_MIN_U32: _VOP3Op_V_MIN_U32,
VOP3Op.V_MAX_U32: _VOP3Op_V_MAX_U32,
VOP3Op.V_LSHLREV_B32: _VOP3Op_V_LSHLREV_B32,
VOP3Op.V_LSHRREV_B32: _VOP3Op_V_LSHRREV_B32,
VOP3Op.V_ASHRREV_I32: _VOP3Op_V_ASHRREV_I32,
VOP3Op.V_AND_B32: _VOP3Op_V_AND_B32,
VOP3Op.V_OR_B32: _VOP3Op_V_OR_B32,
VOP3Op.V_XOR_B32: _VOP3Op_V_XOR_B32,
VOP3Op.V_XNOR_B32: _VOP3Op_V_XNOR_B32,
VOP3Op.V_ADD_NC_U32: _VOP3Op_V_ADD_NC_U32,
VOP3Op.V_SUB_NC_U32: _VOP3Op_V_SUB_NC_U32,
VOP3Op.V_SUBREV_NC_U32: _VOP3Op_V_SUBREV_NC_U32,
VOP3Op.V_FMAC_F32: _VOP3Op_V_FMAC_F32,
VOP3Op.V_CVT_PK_RTZ_F16_F32: _VOP3Op_V_CVT_PK_RTZ_F16_F32,
VOP3Op.V_ADD_F16: _VOP3Op_V_ADD_F16,
VOP3Op.V_SUB_F16: _VOP3Op_V_SUB_F16,
VOP3Op.V_SUBREV_F16: _VOP3Op_V_SUBREV_F16,
VOP3Op.V_MUL_F16: _VOP3Op_V_MUL_F16,
VOP3Op.V_FMAC_F16: _VOP3Op_V_FMAC_F16,
VOP3Op.V_MAX_F16: _VOP3Op_V_MAX_F16,
VOP3Op.V_MIN_F16: _VOP3Op_V_MIN_F16,
VOP3Op.V_LDEXP_F16: _VOP3Op_V_LDEXP_F16,
VOP3Op.V_FMA_DX9_ZERO_F32: _VOP3Op_V_FMA_DX9_ZERO_F32,
VOP3Op.V_MAD_I32_I24: _VOP3Op_V_MAD_I32_I24,
VOP3Op.V_MAD_U32_U24: _VOP3Op_V_MAD_U32_U24,
VOP3Op.V_CUBEID_F32: _VOP3Op_V_CUBEID_F32,
VOP3Op.V_CUBESC_F32: _VOP3Op_V_CUBESC_F32,
VOP3Op.V_CUBETC_F32: _VOP3Op_V_CUBETC_F32,
VOP3Op.V_CUBEMA_F32: _VOP3Op_V_CUBEMA_F32,
VOP3Op.V_BFE_U32: _VOP3Op_V_BFE_U32,
VOP3Op.V_BFE_I32: _VOP3Op_V_BFE_I32,
VOP3Op.V_BFI_B32: _VOP3Op_V_BFI_B32,
VOP3Op.V_FMA_F32: _VOP3Op_V_FMA_F32,
VOP3Op.V_FMA_F64: _VOP3Op_V_FMA_F64,
VOP3Op.V_LERP_U8: _VOP3Op_V_LERP_U8,
VOP3Op.V_ALIGNBIT_B32: _VOP3Op_V_ALIGNBIT_B32,
VOP3Op.V_ALIGNBYTE_B32: _VOP3Op_V_ALIGNBYTE_B32,
VOP3Op.V_MULLIT_F32: _VOP3Op_V_MULLIT_F32,
VOP3Op.V_MIN3_F32: _VOP3Op_V_MIN3_F32,
VOP3Op.V_MIN3_I32: _VOP3Op_V_MIN3_I32,
VOP3Op.V_MIN3_U32: _VOP3Op_V_MIN3_U32,
VOP3Op.V_MAX3_F32: _VOP3Op_V_MAX3_F32,
VOP3Op.V_MAX3_I32: _VOP3Op_V_MAX3_I32,
VOP3Op.V_MAX3_U32: _VOP3Op_V_MAX3_U32,
VOP3Op.V_MED3_F32: _VOP3Op_V_MED3_F32,
VOP3Op.V_MED3_I32: _VOP3Op_V_MED3_I32,
VOP3Op.V_MED3_U32: _VOP3Op_V_MED3_U32,
VOP3Op.V_SAD_U8: _VOP3Op_V_SAD_U8,
VOP3Op.V_SAD_HI_U8: _VOP3Op_V_SAD_HI_U8,
VOP3Op.V_SAD_U16: _VOP3Op_V_SAD_U16,
VOP3Op.V_SAD_U32: _VOP3Op_V_SAD_U32,
VOP3Op.V_CVT_PK_U8_F32: _VOP3Op_V_CVT_PK_U8_F32,
VOP3Op.V_DIV_FIXUP_F32: _VOP3Op_V_DIV_FIXUP_F32,
VOP3Op.V_DIV_FIXUP_F64: _VOP3Op_V_DIV_FIXUP_F64,
VOP3Op.V_DIV_FMAS_F32: _VOP3Op_V_DIV_FMAS_F32,
VOP3Op.V_DIV_FMAS_F64: _VOP3Op_V_DIV_FMAS_F64,
VOP3Op.V_MSAD_U8: _VOP3Op_V_MSAD_U8,
VOP3Op.V_QSAD_PK_U16_U8: _VOP3Op_V_QSAD_PK_U16_U8,
VOP3Op.V_MQSAD_PK_U16_U8: _VOP3Op_V_MQSAD_PK_U16_U8,
VOP3Op.V_MQSAD_U32_U8: _VOP3Op_V_MQSAD_U32_U8,
VOP3Op.V_XOR3_B32: _VOP3Op_V_XOR3_B32,
VOP3Op.V_MAD_U16: _VOP3Op_V_MAD_U16,
VOP3Op.V_XAD_U32: _VOP3Op_V_XAD_U32,
VOP3Op.V_LSHL_ADD_U32: _VOP3Op_V_LSHL_ADD_U32,
VOP3Op.V_ADD_LSHL_U32: _VOP3Op_V_ADD_LSHL_U32,
VOP3Op.V_FMA_F16: _VOP3Op_V_FMA_F16,
VOP3Op.V_MIN3_F16: _VOP3Op_V_MIN3_F16,
VOP3Op.V_MIN3_I16: _VOP3Op_V_MIN3_I16,
VOP3Op.V_MIN3_U16: _VOP3Op_V_MIN3_U16,
VOP3Op.V_MAX3_F16: _VOP3Op_V_MAX3_F16,
VOP3Op.V_MAX3_I16: _VOP3Op_V_MAX3_I16,
VOP3Op.V_MAX3_U16: _VOP3Op_V_MAX3_U16,
VOP3Op.V_MED3_F16: _VOP3Op_V_MED3_F16,
VOP3Op.V_MED3_I16: _VOP3Op_V_MED3_I16,
VOP3Op.V_MED3_U16: _VOP3Op_V_MED3_U16,
VOP3Op.V_MAD_I16: _VOP3Op_V_MAD_I16,
VOP3Op.V_DIV_FIXUP_F16: _VOP3Op_V_DIV_FIXUP_F16,
VOP3Op.V_ADD3_U32: _VOP3Op_V_ADD3_U32,
VOP3Op.V_LSHL_OR_B32: _VOP3Op_V_LSHL_OR_B32,
VOP3Op.V_AND_OR_B32: _VOP3Op_V_AND_OR_B32,
VOP3Op.V_OR3_B32: _VOP3Op_V_OR3_B32,
VOP3Op.V_MAD_U32_U16: _VOP3Op_V_MAD_U32_U16,
VOP3Op.V_MAD_I32_I16: _VOP3Op_V_MAD_I32_I16,
VOP3Op.V_CNDMASK_B16: _VOP3Op_V_CNDMASK_B16,
VOP3Op.V_MAXMIN_F32: _VOP3Op_V_MAXMIN_F32,
VOP3Op.V_MINMAX_F32: _VOP3Op_V_MINMAX_F32,
VOP3Op.V_MAXMIN_F16: _VOP3Op_V_MAXMIN_F16,
VOP3Op.V_MINMAX_F16: _VOP3Op_V_MINMAX_F16,
VOP3Op.V_MAXMIN_U32: _VOP3Op_V_MAXMIN_U32,
VOP3Op.V_MINMAX_U32: _VOP3Op_V_MINMAX_U32,
VOP3Op.V_MAXMIN_I32: _VOP3Op_V_MAXMIN_I32,
VOP3Op.V_MINMAX_I32: _VOP3Op_V_MINMAX_I32,
VOP3Op.V_DOT2_F16_F16: _VOP3Op_V_DOT2_F16_F16,
VOP3Op.V_DOT2_BF16_BF16: _VOP3Op_V_DOT2_BF16_BF16,
VOP3Op.V_ADD_NC_U16: _VOP3Op_V_ADD_NC_U16,
VOP3Op.V_SUB_NC_U16: _VOP3Op_V_SUB_NC_U16,
VOP3Op.V_MUL_LO_U16: _VOP3Op_V_MUL_LO_U16,
VOP3Op.V_CVT_PK_I16_F32: _VOP3Op_V_CVT_PK_I16_F32,
VOP3Op.V_CVT_PK_U16_F32: _VOP3Op_V_CVT_PK_U16_F32,
VOP3Op.V_MAX_U16: _VOP3Op_V_MAX_U16,
VOP3Op.V_MAX_I16: _VOP3Op_V_MAX_I16,
VOP3Op.V_MIN_U16: _VOP3Op_V_MIN_U16,
VOP3Op.V_MIN_I16: _VOP3Op_V_MIN_I16,
VOP3Op.V_ADD_NC_I16: _VOP3Op_V_ADD_NC_I16,
VOP3Op.V_SUB_NC_I16: _VOP3Op_V_SUB_NC_I16,
VOP3Op.V_PACK_B32_F16: _VOP3Op_V_PACK_B32_F16,
VOP3Op.V_CVT_PK_NORM_I16_F16: _VOP3Op_V_CVT_PK_NORM_I16_F16,
VOP3Op.V_CVT_PK_NORM_U16_F16: _VOP3Op_V_CVT_PK_NORM_U16_F16,
VOP3Op.V_LDEXP_F32: _VOP3Op_V_LDEXP_F32,
VOP3Op.V_BFM_B32: _VOP3Op_V_BFM_B32,
VOP3Op.V_BCNT_U32_B32: _VOP3Op_V_BCNT_U32_B32,
VOP3Op.V_CVT_PK_NORM_I16_F32: _VOP3Op_V_CVT_PK_NORM_I16_F32,
VOP3Op.V_CVT_PK_NORM_U16_F32: _VOP3Op_V_CVT_PK_NORM_U16_F32,
VOP3Op.V_CVT_PK_U16_U32: _VOP3Op_V_CVT_PK_U16_U32,
VOP3Op.V_CVT_PK_I16_I32: _VOP3Op_V_CVT_PK_I16_I32,
VOP3Op.V_SUB_NC_I32: _VOP3Op_V_SUB_NC_I32,
VOP3Op.V_ADD_NC_I32: _VOP3Op_V_ADD_NC_I32,
VOP3Op.V_ADD_F64: _VOP3Op_V_ADD_F64,
VOP3Op.V_MUL_F64: _VOP3Op_V_MUL_F64,
VOP3Op.V_MIN_F64: _VOP3Op_V_MIN_F64,
VOP3Op.V_MAX_F64: _VOP3Op_V_MAX_F64,
VOP3Op.V_LDEXP_F64: _VOP3Op_V_LDEXP_F64,
VOP3Op.V_MUL_LO_U32: _VOP3Op_V_MUL_LO_U32,
VOP3Op.V_MUL_HI_U32: _VOP3Op_V_MUL_HI_U32,
VOP3Op.V_MUL_HI_I32: _VOP3Op_V_MUL_HI_I32,
VOP3Op.V_LSHLREV_B16: _VOP3Op_V_LSHLREV_B16,
VOP3Op.V_LSHRREV_B16: _VOP3Op_V_LSHRREV_B16,
VOP3Op.V_ASHRREV_I16: _VOP3Op_V_ASHRREV_I16,
VOP3Op.V_LSHLREV_B64: _VOP3Op_V_LSHLREV_B64,
VOP3Op.V_LSHRREV_B64: _VOP3Op_V_LSHRREV_B64,
VOP3Op.V_ASHRREV_I64: _VOP3Op_V_ASHRREV_I64,
VOP3Op.V_READLANE_B32: _VOP3Op_V_READLANE_B32,
VOP3Op.V_AND_B16: _VOP3Op_V_AND_B16,
VOP3Op.V_OR_B16: _VOP3Op_V_OR_B16,
VOP3Op.V_XOR_B16: _VOP3Op_V_XOR_B16,
}
def _VOP3SDOp_V_ADD_CO_CI_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = 64'U(S0.u32) + 64'U(S1.u32) + VCC.u64[laneId].u64;
# VCC.u64[laneId] = tmp >= 0x100000000ULL ? 1'1U : 1'0U;
# // VCC is an UNSIGNED overflow/carry-out for V_ADD_CO_CI_U32.
# D0.u32 = tmp.u32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg((S0.u32) + (S1.u32) + VCC.u64[laneId])
VCC.u64[laneId] = ((1) if (tmp >= 0x100000000) else (0))
D0.u32 = tmp.u32
# --- end pseudocode ---
def _VOP3SDOp_V_SUB_CO_CI_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = S0.u32 - S1.u32 - VCC.u64[laneId].u32;
# VCC.u64[laneId] = 64'U(S1.u32) + VCC.u64[laneId].u64 > 64'U(S0.u32) ? 1'1U : 1'0U;
# // VCC is an UNSIGNED overflow/carry-out for V_SUB_CO_CI_U32.
# D0.u32 = tmp.u32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(S0.u32 - S1.u32 - VCC.u64[laneId])
VCC.u64[laneId] = ((1) if ((S1.u32) + VCC.u64[laneId] > (S0.u32)) else (0))
D0.u32 = tmp.u32
# --- end pseudocode ---
def _VOP3SDOp_V_SUBREV_CO_CI_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = S1.u32 - S0.u32 - VCC.u64[laneId].u32;
# VCC.u64[laneId] = 64'U(S0.u32) + VCC.u64[laneId].u64 > 64'U(S1.u32) ? 1'1U : 1'0U;
# // VCC is an UNSIGNED overflow/carry-out for V_SUB_CO_CI_U32.
# D0.u32 = tmp.u32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(S1.u32 - S0.u32 - VCC.u64[laneId])
VCC.u64[laneId] = ((1) if ((S0.u32) + VCC.u64[laneId] > (S1.u32)) else (0))
D0.u32 = tmp.u32
# --- end pseudocode ---
def _VOP3SDOp_V_DIV_SCALE_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# VCC = 0x0LL;
# if ((64'F(S2.f32) == 0.0) || (64'F(S1.f32) == 0.0)) then
# D0.f32 = NAN.f32
# elsif exponent(S2.f32) - exponent(S1.f32) >= 96 then
# // N/D near MAX_FLOAT_F32
# VCC = 0x1LL;
# if S0.f32 == S1.f32 then
# // Only scale the denominator
# D0.f32 = ldexp(S0.f32, 64)
# endif
# elsif S1.f32 == DENORM.f32 then
# D0.f32 = ldexp(S0.f32, 64)
# elsif ((1.0 / 64'F(S1.f32) == DENORM.f64) && (S2.f32 / S1.f32 == DENORM.f32)) then
# VCC = 0x1LL;
# if S0.f32 == S1.f32 then
# // Only scale the denominator
# D0.f32 = ldexp(S0.f32, 64)
# endif
# elsif 1.0 / 64'F(S1.f32) == DENORM.f64 then
# D0.f32 = ldexp(S0.f32, -64)
# elsif S2.f32 / S1.f32 == DENORM.f32 then
# VCC = 0x1LL;
# if S0.f32 == S2.f32 then
# // Only scale the numerator
# D0.f32 = ldexp(S0.f32, 64)
# endif
# elsif exponent(S2.f32) <= 23 then
# // Numerator is tiny
# D0.f32 = ldexp(S0.f32, 64)
# endif
D0._val = S0._val
# --- compiled pseudocode ---
# V_DIV_SCALE sets VCC bit for the lane if scaling is needed
VCC.u64[laneId] = 0
if ((F(S2.f32) == 0.0) or (F(S1.f32) == 0.0)):
VCC.u64[laneId] = 1; D0.f32 = float("nan")
elif exponent(S2.f32) - exponent(S1.f32) >= 96:
VCC.u64[laneId] = 1
if S0.f32 == S1.f32:
D0.f32 = ldexp(S0.f32, 64)
elif S1.f32 == DENORM.f32:
D0.f32 = float("nan")
elif ((1.0 / F(S1.f32) == DENORM.f64) and (S2.f32 / S1.f32 == DENORM.f32)):
VCC.u64[laneId] = 1
if S0.f32 == S1.f32:
D0.f32 = ldexp(S0.f32, 64)
elif 1.0 / F(S1.f32) == DENORM.f64:
D0.f32 = ldexp(S0.f32, -64)
elif S2.f32 / S1.f32 == DENORM.f32:
VCC.u64[laneId] = 1
if S0.f32 == S2.f32:
D0.f32 = ldexp(S0.f32, 64)
elif exponent(S2.f32) <= 23:
VCC.u64[laneId] = 1; D0.f32 = ldexp(S0.f32, 64)
# --- end pseudocode ---
def _VOP3SDOp_V_DIV_SCALE_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# VCC = 0x0LL;
# if ((S2.f64 == 0.0) || (S1.f64 == 0.0)) then
# D0.f64 = NAN.f64
# elsif exponent(S2.f64) - exponent(S1.f64) >= 768 then
# // N/D near MAX_FLOAT_F64
# VCC = 0x1LL;
# if S0.f64 == S1.f64 then
# // Only scale the denominator
# D0.f64 = ldexp(S0.f64, 128)
# endif
# elsif S1.f64 == DENORM.f64 then
# D0.f64 = ldexp(S0.f64, 128)
# elsif ((1.0 / S1.f64 == DENORM.f64) && (S2.f64 / S1.f64 == DENORM.f64)) then
# VCC = 0x1LL;
# if S0.f64 == S1.f64 then
# // Only scale the denominator
# D0.f64 = ldexp(S0.f64, 128)
# endif
# elsif 1.0 / S1.f64 == DENORM.f64 then
# D0.f64 = ldexp(S0.f64, -128)
# elsif S2.f64 / S1.f64 == DENORM.f64 then
# VCC = 0x1LL;
# if S0.f64 == S2.f64 then
# // Only scale the numerator
# D0.f64 = ldexp(S0.f64, 128)
# endif
# elsif exponent(S2.f64) <= 53 then
# // Numerator is tiny
# D0.f64 = ldexp(S0.f64, 128)
# endif
D0._val = S0._val
# --- compiled pseudocode ---
# V_DIV_SCALE sets VCC bit for the lane if scaling is needed
VCC.u64[laneId] = 0
if ((S2.f64 == 0.0) or (S1.f64 == 0.0)):
VCC.u64[laneId] = 1; D0.f64 = float("nan")
elif exponent(S2.f64) - exponent(S1.f64) >= 768:
VCC.u64[laneId] = 1
if S0.f64 == S1.f64:
D0.f64 = ldexp(S0.f64, 128)
elif S1.f64 == DENORM.f64:
D0.f64 = float("nan")
elif ((1.0 / S1.f64 == DENORM.f64) and (S2.f64 / S1.f64 == DENORM.f64)):
VCC.u64[laneId] = 1
if S0.f64 == S1.f64:
D0.f64 = ldexp(S0.f64, 128)
elif 1.0 / S1.f64 == DENORM.f64:
D0.f64 = ldexp(S0.f64, -128)
elif S2.f64 / S1.f64 == DENORM.f64:
VCC.u64[laneId] = 1
if S0.f64 == S2.f64:
D0.f64 = ldexp(S0.f64, 128)
elif exponent(S2.f64) <= 53:
VCC.u64[laneId] = 1; D0.f64 = ldexp(S0.f64, 128)
# --- end pseudocode ---
def _VOP3SDOp_V_MAD_U64_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# { D1.u1, D0.u64 } = 65'B(65'U(S0.u32) * 65'U(S1.u32) + 65'U(S2.u64))
D1 = Reg(0)
# --- compiled pseudocode ---
_full = ((S0.u32) * (S1.u32) + (S2.u64))
D0.u64 = int(_full) & 0xffffffffffffffff
D1 = Reg((int(_full) >> 64) & 1)
# --- end pseudocode ---
def _VOP3SDOp_V_MAD_I64_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# { D1.i1, D0.i64 } = 65'B(65'I(S0.i32) * 65'I(S1.i32) + 65'I(S2.i64))
D1 = Reg(0)
# --- compiled pseudocode ---
_full = ((S0.i32) * (S1.i32) + (S2.i64))
D0.u64 = int(_full) & 0xffffffffffffffff
D1 = Reg((int(_full) >> 64) & 1)
# --- end pseudocode ---
def _VOP3SDOp_V_ADD_CO_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = 64'U(S0.u32) + 64'U(S1.u32);
# VCC.u64[laneId] = tmp >= 0x100000000ULL ? 1'1U : 1'0U;
# // VCC is an UNSIGNED overflow/carry-out for V_ADD_CO_CI_U32.
# D0.u32 = tmp.u32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg((S0.u32) + (S1.u32))
VCC.u64[laneId] = ((1) if (tmp >= 0x100000000) else (0))
D0.u32 = tmp.u32
# --- end pseudocode ---
def _VOP3SDOp_V_SUB_CO_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = S0.u32 - S1.u32;
# VCC.u64[laneId] = S1.u32 > S0.u32 ? 1'1U : 1'0U;
# // VCC is an UNSIGNED overflow/carry-out for V_SUB_CO_CI_U32.
# D0.u32 = tmp.u32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(S0.u32 - S1.u32)
VCC.u64[laneId] = ((1) if (S1.u32 > S0.u32) else (0))
D0.u32 = tmp.u32
# --- end pseudocode ---
def _VOP3SDOp_V_SUBREV_CO_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = S1.u32 - S0.u32;
# VCC.u64[laneId] = S0.u32 > S1.u32 ? 1'1U : 1'0U;
# // VCC is an UNSIGNED overflow/carry-out for V_SUB_CO_CI_U32.
# D0.u32 = tmp.u32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(S1.u32 - S0.u32)
VCC.u64[laneId] = ((1) if (S0.u32 > S1.u32) else (0))
D0.u32 = tmp.u32
# --- end pseudocode ---
VOP3SDOp_FUNCTIONS = {
VOP3SDOp.V_ADD_CO_CI_U32: _VOP3SDOp_V_ADD_CO_CI_U32,
VOP3SDOp.V_SUB_CO_CI_U32: _VOP3SDOp_V_SUB_CO_CI_U32,
VOP3SDOp.V_SUBREV_CO_CI_U32: _VOP3SDOp_V_SUBREV_CO_CI_U32,
VOP3SDOp.V_DIV_SCALE_F32: _VOP3SDOp_V_DIV_SCALE_F32,
VOP3SDOp.V_DIV_SCALE_F64: _VOP3SDOp_V_DIV_SCALE_F64,
VOP3SDOp.V_MAD_U64_U32: _VOP3SDOp_V_MAD_U64_U32,
VOP3SDOp.V_MAD_I64_I32: _VOP3SDOp_V_MAD_I64_I32,
VOP3SDOp.V_ADD_CO_U32: _VOP3SDOp_V_ADD_CO_U32,
VOP3SDOp.V_SUB_CO_U32: _VOP3SDOp_V_SUB_CO_U32,
VOP3SDOp.V_SUBREV_CO_U32: _VOP3SDOp_V_SUBREV_CO_U32,
}
def _VOP3POp_V_PK_MAD_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp[31 : 16].i16 = S0[31 : 16].i16 * S1[31 : 16].i16 + S2[31 : 16].i16;
# tmp[15 : 0].i16 = S0[15 : 0].i16 * S1[15 : 0].i16 + S2[15 : 0].i16;
# D0.b32 = tmp.b32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[31 : 16].i16 = S0[31 : 16].i16 * S1[31 : 16].i16 + S2[31 : 16].i16
tmp[15 : 0].i16 = S0[15 : 0].i16 * S1[15 : 0].i16 + S2[15 : 0].i16
D0.b32 = tmp.b32
# --- end pseudocode ---
def _VOP3POp_V_PK_MUL_LO_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp[31 : 16].u16 = S0[31 : 16].u16 * S1[31 : 16].u16;
# tmp[15 : 0].u16 = S0[15 : 0].u16 * S1[15 : 0].u16;
# D0.b32 = tmp.b32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[31 : 16].u16 = S0[31 : 16].u16 * S1[31 : 16].u16
tmp[15 : 0].u16 = S0[15 : 0].u16 * S1[15 : 0].u16
D0.b32 = tmp.b32
# --- end pseudocode ---
def _VOP3POp_V_PK_ADD_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp[31 : 16].i16 = S0[31 : 16].i16 + S1[31 : 16].i16;
# tmp[15 : 0].i16 = S0[15 : 0].i16 + S1[15 : 0].i16;
# D0.b32 = tmp.b32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[31 : 16].i16 = S0[31 : 16].i16 + S1[31 : 16].i16
tmp[15 : 0].i16 = S0[15 : 0].i16 + S1[15 : 0].i16
D0.b32 = tmp.b32
# --- end pseudocode ---
def _VOP3POp_V_PK_SUB_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp[31 : 16].i16 = S0[31 : 16].i16 - S1[31 : 16].i16;
# tmp[15 : 0].i16 = S0[15 : 0].i16 - S1[15 : 0].i16;
# D0.b32 = tmp.b32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[31 : 16].i16 = S0[31 : 16].i16 - S1[31 : 16].i16
tmp[15 : 0].i16 = S0[15 : 0].i16 - S1[15 : 0].i16
D0.b32 = tmp.b32
# --- end pseudocode ---
def _VOP3POp_V_PK_LSHLREV_B16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp[31 : 16].u16 = (S1[31 : 16].u16 << S0.u32[19 : 16].u32);
# tmp[15 : 0].u16 = (S1[15 : 0].u16 << S0.u32[3 : 0].u32);
# D0.b32 = tmp.b32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[31 : 16].u16 = (S1[31 : 16].u16 << S0.u32[19 : 16].u32)
tmp[15 : 0].u16 = (S1[15 : 0].u16 << S0.u32[3 : 0].u32)
D0.b32 = tmp.b32
# --- end pseudocode ---
def _VOP3POp_V_PK_LSHRREV_B16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp[31 : 16].u16 = (S1[31 : 16].u16 >> S0.u32[19 : 16].u32);
# tmp[15 : 0].u16 = (S1[15 : 0].u16 >> S0.u32[3 : 0].u32);
# D0.b32 = tmp.b32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[31 : 16].u16 = (S1[31 : 16].u16 >> S0.u32[19 : 16].u32)
tmp[15 : 0].u16 = (S1[15 : 0].u16 >> S0.u32[3 : 0].u32)
D0.b32 = tmp.b32
# --- end pseudocode ---
def _VOP3POp_V_PK_ASHRREV_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp[31 : 16].i16 = (S1[31 : 16].i16 >> S0.u32[19 : 16].u32);
# tmp[15 : 0].i16 = (S1[15 : 0].i16 >> S0.u32[3 : 0].u32);
# D0.b32 = tmp.b32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[31 : 16].i16 = (S1[31 : 16].i16 >> S0.u32[19 : 16].u32)
tmp[15 : 0].i16 = (S1[15 : 0].i16 >> S0.u32[3 : 0].u32)
D0.b32 = tmp.b32
# --- end pseudocode ---
def _VOP3POp_V_PK_MAX_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp[31 : 16].i16 = S0[31 : 16].i16 >= S1[31 : 16].i16 ? S0[31 : 16].i16 : S1[31 : 16].i16;
# tmp[15 : 0].i16 = S0[15 : 0].i16 >= S1[15 : 0].i16 ? S0[15 : 0].i16 : S1[15 : 0].i16;
# D0.b32 = tmp.b32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[31 : 16].i16 = ((S0[31 : 16].i16) if (S0[31 : 16].i16 >= S1[31 : 16].i16) else (S1[31 : 16].i16))
tmp[15 : 0].i16 = ((S0[15 : 0].i16) if (S0[15 : 0].i16 >= S1[15 : 0].i16) else (S1[15 : 0].i16))
D0.b32 = tmp.b32
# --- end pseudocode ---
def _VOP3POp_V_PK_MIN_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp[31 : 16].i16 = S0[31 : 16].i16 < S1[31 : 16].i16 ? S0[31 : 16].i16 : S1[31 : 16].i16;
# tmp[15 : 0].i16 = S0[15 : 0].i16 < S1[15 : 0].i16 ? S0[15 : 0].i16 : S1[15 : 0].i16;
# D0.b32 = tmp.b32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[31 : 16].i16 = ((S0[31 : 16].i16) if (S0[31 : 16].i16 < S1[31 : 16].i16) else (S1[31 : 16].i16))
tmp[15 : 0].i16 = ((S0[15 : 0].i16) if (S0[15 : 0].i16 < S1[15 : 0].i16) else (S1[15 : 0].i16))
D0.b32 = tmp.b32
# --- end pseudocode ---
def _VOP3POp_V_PK_MAD_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp[31 : 16].u16 = S0[31 : 16].u16 * S1[31 : 16].u16 + S2[31 : 16].u16;
# tmp[15 : 0].u16 = S0[15 : 0].u16 * S1[15 : 0].u16 + S2[15 : 0].u16;
# D0.b32 = tmp.b32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[31 : 16].u16 = S0[31 : 16].u16 * S1[31 : 16].u16 + S2[31 : 16].u16
tmp[15 : 0].u16 = S0[15 : 0].u16 * S1[15 : 0].u16 + S2[15 : 0].u16
D0.b32 = tmp.b32
# --- end pseudocode ---
def _VOP3POp_V_PK_ADD_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp[31 : 16].u16 = S0[31 : 16].u16 + S1[31 : 16].u16;
# tmp[15 : 0].u16 = S0[15 : 0].u16 + S1[15 : 0].u16;
# D0.b32 = tmp.b32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[31 : 16].u16 = S0[31 : 16].u16 + S1[31 : 16].u16
tmp[15 : 0].u16 = S0[15 : 0].u16 + S1[15 : 0].u16
D0.b32 = tmp.b32
# --- end pseudocode ---
def _VOP3POp_V_PK_SUB_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp[31 : 16].u16 = S0[31 : 16].u16 - S1[31 : 16].u16;
# tmp[15 : 0].u16 = S0[15 : 0].u16 - S1[15 : 0].u16;
# D0.b32 = tmp.b32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[31 : 16].u16 = S0[31 : 16].u16 - S1[31 : 16].u16
tmp[15 : 0].u16 = S0[15 : 0].u16 - S1[15 : 0].u16
D0.b32 = tmp.b32
# --- end pseudocode ---
def _VOP3POp_V_PK_MAX_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp[31 : 16].u16 = S0[31 : 16].u16 >= S1[31 : 16].u16 ? S0[31 : 16].u16 : S1[31 : 16].u16;
# tmp[15 : 0].u16 = S0[15 : 0].u16 >= S1[15 : 0].u16 ? S0[15 : 0].u16 : S1[15 : 0].u16;
# D0.b32 = tmp.b32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[31 : 16].u16 = ((S0[31 : 16].u16) if (S0[31 : 16].u16 >= S1[31 : 16].u16) else (S1[31 : 16].u16))
tmp[15 : 0].u16 = ((S0[15 : 0].u16) if (S0[15 : 0].u16 >= S1[15 : 0].u16) else (S1[15 : 0].u16))
D0.b32 = tmp.b32
# --- end pseudocode ---
def _VOP3POp_V_PK_MIN_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp[31 : 16].u16 = S0[31 : 16].u16 < S1[31 : 16].u16 ? S0[31 : 16].u16 : S1[31 : 16].u16;
# tmp[15 : 0].u16 = S0[15 : 0].u16 < S1[15 : 0].u16 ? S0[15 : 0].u16 : S1[15 : 0].u16;
# D0.b32 = tmp.b32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[31 : 16].u16 = ((S0[31 : 16].u16) if (S0[31 : 16].u16 < S1[31 : 16].u16) else (S1[31 : 16].u16))
tmp[15 : 0].u16 = ((S0[15 : 0].u16) if (S0[15 : 0].u16 < S1[15 : 0].u16) else (S1[15 : 0].u16))
D0.b32 = tmp.b32
# --- end pseudocode ---
def _VOP3POp_V_PK_FMA_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# declare tmp : 32'B;
# tmp[31 : 16].f16 = fma(S0[31 : 16].f16, S1[31 : 16].f16, S2[31 : 16].f16);
# tmp[15 : 0].f16 = fma(S0[15 : 0].f16, S1[15 : 0].f16, S2[15 : 0].f16);
# D0.b32 = tmp
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[31 : 16].f16 = fma(S0[31 : 16].f16, S1[31 : 16].f16, S2[31 : 16].f16)
tmp[15 : 0].f16 = fma(S0[15 : 0].f16, S1[15 : 0].f16, S2[15 : 0].f16)
D0.b32 = tmp
# --- end pseudocode ---
def _VOP3POp_V_PK_ADD_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp[31 : 16].f16 = S0[31 : 16].f16 + S1[31 : 16].f16;
# tmp[15 : 0].f16 = S0[15 : 0].f16 + S1[15 : 0].f16;
# D0.b32 = tmp.b32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[31 : 16].f16 = S0[31 : 16].f16 + S1[31 : 16].f16
tmp[15 : 0].f16 = S0[15 : 0].f16 + S1[15 : 0].f16
D0.b32 = tmp.b32
# --- end pseudocode ---
def _VOP3POp_V_PK_MUL_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp[31 : 16].f16 = S0[31 : 16].f16 * S1[31 : 16].f16;
# tmp[15 : 0].f16 = S0[15 : 0].f16 * S1[15 : 0].f16;
# D0.b32 = tmp.b32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[31 : 16].f16 = S0[31 : 16].f16 * S1[31 : 16].f16
tmp[15 : 0].f16 = S0[15 : 0].f16 * S1[15 : 0].f16
D0.b32 = tmp.b32
# --- end pseudocode ---
def _VOP3POp_V_PK_MIN_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp[31 : 16].f16 = v_min_f16(S0[31 : 16].f16, S1[31 : 16].f16);
# tmp[15 : 0].f16 = v_min_f16(S0[15 : 0].f16, S1[15 : 0].f16);
# D0.b32 = tmp.b32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[31 : 16].f16 = v_min_f16(S0[31 : 16].f16, S1[31 : 16].f16)
tmp[15 : 0].f16 = v_min_f16(S0[15 : 0].f16, S1[15 : 0].f16)
D0.b32 = tmp.b32
# --- end pseudocode ---
def _VOP3POp_V_PK_MAX_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp[31 : 16].f16 = v_max_f16(S0[31 : 16].f16, S1[31 : 16].f16);
# tmp[15 : 0].f16 = v_max_f16(S0[15 : 0].f16, S1[15 : 0].f16);
# D0.b32 = tmp.b32
tmp = Reg(0)
# --- compiled pseudocode ---
tmp[31 : 16].f16 = v_max_f16(S0[31 : 16].f16, S1[31 : 16].f16)
tmp[15 : 0].f16 = v_max_f16(S0[15 : 0].f16, S1[15 : 0].f16)
D0.b32 = tmp.b32
# --- end pseudocode ---
def _VOP3POp_V_DOT2_F32_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = S2.f32;
# tmp += f16_to_f32(S0[15 : 0].f16) * f16_to_f32(S1[15 : 0].f16);
# tmp += f16_to_f32(S0[31 : 16].f16) * f16_to_f32(S1[31 : 16].f16);
# D0.f32 = tmp
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(S2.f32)
tmp += f16_to_f32(S0[15 : 0].f16) * f16_to_f32(S1[15 : 0].f16)
tmp += f16_to_f32(S0[31 : 16].f16) * f16_to_f32(S1[31 : 16].f16)
D0.f32 = tmp
# --- end pseudocode ---
def _VOP3POp_V_DOT4_U32_U8(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = S2.u32;
# tmp += u8_to_u32(S0[7 : 0].u8) * u8_to_u32(S1[7 : 0].u8);
# tmp += u8_to_u32(S0[15 : 8].u8) * u8_to_u32(S1[15 : 8].u8);
# tmp += u8_to_u32(S0[23 : 16].u8) * u8_to_u32(S1[23 : 16].u8);
# tmp += u8_to_u32(S0[31 : 24].u8) * u8_to_u32(S1[31 : 24].u8);
# D0.u32 = tmp
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(S2.u32)
tmp += u8_to_u32(S0[7 : 0].u8) * u8_to_u32(S1[7 : 0].u8)
tmp += u8_to_u32(S0[15 : 8].u8) * u8_to_u32(S1[15 : 8].u8)
tmp += u8_to_u32(S0[23 : 16].u8) * u8_to_u32(S1[23 : 16].u8)
tmp += u8_to_u32(S0[31 : 24].u8) * u8_to_u32(S1[31 : 24].u8)
D0.u32 = tmp
# --- end pseudocode ---
def _VOP3POp_V_DOT8_U32_U4(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = S2.u32;
# tmp += u4_to_u32(S0[3 : 0].u4) * u4_to_u32(S1[3 : 0].u4);
# tmp += u4_to_u32(S0[7 : 4].u4) * u4_to_u32(S1[7 : 4].u4);
# tmp += u4_to_u32(S0[11 : 8].u4) * u4_to_u32(S1[11 : 8].u4);
# tmp += u4_to_u32(S0[15 : 12].u4) * u4_to_u32(S1[15 : 12].u4);
# tmp += u4_to_u32(S0[19 : 16].u4) * u4_to_u32(S1[19 : 16].u4);
# tmp += u4_to_u32(S0[23 : 20].u4) * u4_to_u32(S1[23 : 20].u4);
# tmp += u4_to_u32(S0[27 : 24].u4) * u4_to_u32(S1[27 : 24].u4);
# tmp += u4_to_u32(S0[31 : 28].u4) * u4_to_u32(S1[31 : 28].u4);
# D0.u32 = tmp
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(S2.u32)
tmp += u4_to_u32(S0[3 : 0].u4) * u4_to_u32(S1[3 : 0].u4)
tmp += u4_to_u32(S0[7 : 4].u4) * u4_to_u32(S1[7 : 4].u4)
tmp += u4_to_u32(S0[11 : 8].u4) * u4_to_u32(S1[11 : 8].u4)
tmp += u4_to_u32(S0[15 : 12].u4) * u4_to_u32(S1[15 : 12].u4)
tmp += u4_to_u32(S0[19 : 16].u4) * u4_to_u32(S1[19 : 16].u4)
tmp += u4_to_u32(S0[23 : 20].u4) * u4_to_u32(S1[23 : 20].u4)
tmp += u4_to_u32(S0[27 : 24].u4) * u4_to_u32(S1[27 : 24].u4)
tmp += u4_to_u32(S0[31 : 28].u4) * u4_to_u32(S1[31 : 28].u4)
D0.u32 = tmp
# --- end pseudocode ---
def _VOP3POp_V_DOT2_F32_BF16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# tmp = S2.f32;
# tmp += bf16_to_f32(S0[15 : 0].bf16) * bf16_to_f32(S1[15 : 0].bf16);
# tmp += bf16_to_f32(S0[31 : 16].bf16) * bf16_to_f32(S1[31 : 16].bf16);
# D0.f32 = tmp
tmp = Reg(0)
# --- compiled pseudocode ---
tmp = Reg(S2.f32)
tmp += bf16_to_f32(S0[15 : 0].bf16) * bf16_to_f32(S1[15 : 0].bf16)
tmp += bf16_to_f32(S0[31 : 16].bf16) * bf16_to_f32(S1[31 : 16].bf16)
D0.f32 = tmp
# --- end pseudocode ---
VOP3POp_FUNCTIONS = {
VOP3POp.V_PK_MAD_I16: _VOP3POp_V_PK_MAD_I16,
VOP3POp.V_PK_MUL_LO_U16: _VOP3POp_V_PK_MUL_LO_U16,
VOP3POp.V_PK_ADD_I16: _VOP3POp_V_PK_ADD_I16,
VOP3POp.V_PK_SUB_I16: _VOP3POp_V_PK_SUB_I16,
VOP3POp.V_PK_LSHLREV_B16: _VOP3POp_V_PK_LSHLREV_B16,
VOP3POp.V_PK_LSHRREV_B16: _VOP3POp_V_PK_LSHRREV_B16,
VOP3POp.V_PK_ASHRREV_I16: _VOP3POp_V_PK_ASHRREV_I16,
VOP3POp.V_PK_MAX_I16: _VOP3POp_V_PK_MAX_I16,
VOP3POp.V_PK_MIN_I16: _VOP3POp_V_PK_MIN_I16,
VOP3POp.V_PK_MAD_U16: _VOP3POp_V_PK_MAD_U16,
VOP3POp.V_PK_ADD_U16: _VOP3POp_V_PK_ADD_U16,
VOP3POp.V_PK_SUB_U16: _VOP3POp_V_PK_SUB_U16,
VOP3POp.V_PK_MAX_U16: _VOP3POp_V_PK_MAX_U16,
VOP3POp.V_PK_MIN_U16: _VOP3POp_V_PK_MIN_U16,
VOP3POp.V_PK_FMA_F16: _VOP3POp_V_PK_FMA_F16,
VOP3POp.V_PK_ADD_F16: _VOP3POp_V_PK_ADD_F16,
VOP3POp.V_PK_MUL_F16: _VOP3POp_V_PK_MUL_F16,
VOP3POp.V_PK_MIN_F16: _VOP3POp_V_PK_MIN_F16,
VOP3POp.V_PK_MAX_F16: _VOP3POp_V_PK_MAX_F16,
VOP3POp.V_DOT2_F32_F16: _VOP3POp_V_DOT2_F32_F16,
VOP3POp.V_DOT4_U32_U8: _VOP3POp_V_DOT4_U32_U8,
VOP3POp.V_DOT8_U32_U4: _VOP3POp_V_DOT8_U32_U4,
VOP3POp.V_DOT2_F32_BF16: _VOP3POp_V_DOT2_F32_BF16,
}
def _VOPCOp_V_CMP_F_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 0. Store the result into VCC or a scalar register.
# D0.u64[laneId] = 1'0U;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = 0
# --- end pseudocode ---
def _VOPCOp_V_CMP_LT_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is less than the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.f16 < S1.f16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f16 < S1.f16
# --- end pseudocode ---
def _VOPCOp_V_CMP_EQ_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.f16 == S1.f16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f16 == S1.f16
# --- end pseudocode ---
def _VOPCOp_V_CMP_LE_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.f16 <= S1.f16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f16 <= S1.f16
# --- end pseudocode ---
def _VOPCOp_V_CMP_GT_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is greater than the second input. Store the result into VCC
# D0.u64[laneId] = S0.f16 > S1.f16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f16 > S1.f16
# --- end pseudocode ---
def _VOPCOp_V_CMP_LG_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.f16 <> S1.f16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f16 != S1.f16
# --- end pseudocode ---
def _VOPCOp_V_CMP_GE_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.f16 >= S1.f16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f16 >= S1.f16
# --- end pseudocode ---
def _VOPCOp_V_CMP_O_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is orderable to the second input. Store the result into VCC
# D0.u64[laneId] = (!isNAN(64'F(S0.f16)) && !isNAN(64'F(S1.f16)));
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = ( not isNAN(F(S0.f16)) and not isNAN(F(S1.f16)))
# --- end pseudocode ---
def _VOPCOp_V_CMP_U_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# VCC or a scalar register.
# D0.u64[laneId] = (isNAN(64'F(S0.f16)) || isNAN(64'F(S1.f16)));
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = (isNAN(F(S0.f16)) or isNAN(F(S1.f16)))
# --- end pseudocode ---
def _VOPCOp_V_CMP_NGE_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = !(S0.f16 >= S1.f16);
# // With NAN inputs this is not the same operation as <
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f16 >= S1.f16)
# --- end pseudocode ---
def _VOPCOp_V_CMP_NLG_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = !(S0.f16 <> S1.f16);
# // With NAN inputs this is not the same operation as ==
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f16 != S1.f16)
# --- end pseudocode ---
def _VOPCOp_V_CMP_NGT_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# VCC or a scalar register.
# D0.u64[laneId] = !(S0.f16 > S1.f16);
# // With NAN inputs this is not the same operation as <=
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f16 > S1.f16)
# --- end pseudocode ---
def _VOPCOp_V_CMP_NLE_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = !(S0.f16 <= S1.f16);
# // With NAN inputs this is not the same operation as >
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f16 <= S1.f16)
# --- end pseudocode ---
def _VOPCOp_V_CMP_NEQ_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is not equal to the second input. Store the result into VCC
# D0.u64[laneId] = !(S0.f16 == S1.f16);
# // With NAN inputs this is not the same operation as !=
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f16 == S1.f16)
# --- end pseudocode ---
def _VOPCOp_V_CMP_NLT_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is not less than the second input. Store the result into VCC
# D0.u64[laneId] = !(S0.f16 < S1.f16);
# // With NAN inputs this is not the same operation as >=
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f16 < S1.f16)
# --- end pseudocode ---
def _VOPCOp_V_CMP_T_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1. Store the result into VCC or a scalar register.
# D0.u64[laneId] = 1'1U;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = 1
# --- end pseudocode ---
def _VOPCOp_V_CMP_F_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 0. Store the result into VCC or a scalar register.
# D0.u64[laneId] = 1'0U;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = 0
# --- end pseudocode ---
def _VOPCOp_V_CMP_LT_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is less than the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.f32 < S1.f32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f32 < S1.f32
# --- end pseudocode ---
def _VOPCOp_V_CMP_EQ_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.f32 == S1.f32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f32 == S1.f32
# --- end pseudocode ---
def _VOPCOp_V_CMP_LE_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.f32 <= S1.f32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f32 <= S1.f32
# --- end pseudocode ---
def _VOPCOp_V_CMP_GT_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is greater than the second input. Store the result into VCC
# D0.u64[laneId] = S0.f32 > S1.f32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f32 > S1.f32
# --- end pseudocode ---
def _VOPCOp_V_CMP_LG_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.f32 <> S1.f32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f32 != S1.f32
# --- end pseudocode ---
def _VOPCOp_V_CMP_GE_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.f32 >= S1.f32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f32 >= S1.f32
# --- end pseudocode ---
def _VOPCOp_V_CMP_O_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is orderable to the second input. Store the result into VCC
# D0.u64[laneId] = (!isNAN(64'F(S0.f32)) && !isNAN(64'F(S1.f32)));
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = ( not isNAN(F(S0.f32)) and not isNAN(F(S1.f32)))
# --- end pseudocode ---
def _VOPCOp_V_CMP_U_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# VCC or a scalar register.
# D0.u64[laneId] = (isNAN(64'F(S0.f32)) || isNAN(64'F(S1.f32)));
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = (isNAN(F(S0.f32)) or isNAN(F(S1.f32)))
# --- end pseudocode ---
def _VOPCOp_V_CMP_NGE_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = !(S0.f32 >= S1.f32);
# // With NAN inputs this is not the same operation as <
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f32 >= S1.f32)
# --- end pseudocode ---
def _VOPCOp_V_CMP_NLG_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = !(S0.f32 <> S1.f32);
# // With NAN inputs this is not the same operation as ==
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f32 != S1.f32)
# --- end pseudocode ---
def _VOPCOp_V_CMP_NGT_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# VCC or a scalar register.
# D0.u64[laneId] = !(S0.f32 > S1.f32);
# // With NAN inputs this is not the same operation as <=
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f32 > S1.f32)
# --- end pseudocode ---
def _VOPCOp_V_CMP_NLE_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = !(S0.f32 <= S1.f32);
# // With NAN inputs this is not the same operation as >
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f32 <= S1.f32)
# --- end pseudocode ---
def _VOPCOp_V_CMP_NEQ_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is not equal to the second input. Store the result into VCC
# D0.u64[laneId] = !(S0.f32 == S1.f32);
# // With NAN inputs this is not the same operation as !=
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f32 == S1.f32)
# --- end pseudocode ---
def _VOPCOp_V_CMP_NLT_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is not less than the second input. Store the result into VCC
# D0.u64[laneId] = !(S0.f32 < S1.f32);
# // With NAN inputs this is not the same operation as >=
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f32 < S1.f32)
# --- end pseudocode ---
def _VOPCOp_V_CMP_T_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1. Store the result into VCC or a scalar register.
# D0.u64[laneId] = 1'1U;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = 1
# --- end pseudocode ---
def _VOPCOp_V_CMP_F_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 0. Store the result into VCC or a scalar register.
# D0.u64[laneId] = 1'0U;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = 0
# --- end pseudocode ---
def _VOPCOp_V_CMP_LT_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is less than the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.f64 < S1.f64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f64 < S1.f64
# --- end pseudocode ---
def _VOPCOp_V_CMP_EQ_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.f64 == S1.f64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f64 == S1.f64
# --- end pseudocode ---
def _VOPCOp_V_CMP_LE_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.f64 <= S1.f64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f64 <= S1.f64
# --- end pseudocode ---
def _VOPCOp_V_CMP_GT_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is greater than the second input. Store the result into VCC
# D0.u64[laneId] = S0.f64 > S1.f64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f64 > S1.f64
# --- end pseudocode ---
def _VOPCOp_V_CMP_LG_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.f64 <> S1.f64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f64 != S1.f64
# --- end pseudocode ---
def _VOPCOp_V_CMP_GE_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.f64 >= S1.f64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.f64 >= S1.f64
# --- end pseudocode ---
def _VOPCOp_V_CMP_O_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is orderable to the second input. Store the result into VCC
# D0.u64[laneId] = (!isNAN(S0.f64) && !isNAN(S1.f64));
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = ( not isNAN(S0.f64) and not isNAN(S1.f64))
# --- end pseudocode ---
def _VOPCOp_V_CMP_U_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# VCC or a scalar register.
# D0.u64[laneId] = (isNAN(S0.f64) || isNAN(S1.f64));
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = (isNAN(S0.f64) or isNAN(S1.f64))
# --- end pseudocode ---
def _VOPCOp_V_CMP_NGE_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = !(S0.f64 >= S1.f64);
# // With NAN inputs this is not the same operation as <
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f64 >= S1.f64)
# --- end pseudocode ---
def _VOPCOp_V_CMP_NLG_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = !(S0.f64 <> S1.f64);
# // With NAN inputs this is not the same operation as ==
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f64 != S1.f64)
# --- end pseudocode ---
def _VOPCOp_V_CMP_NGT_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# VCC or a scalar register.
# D0.u64[laneId] = !(S0.f64 > S1.f64);
# // With NAN inputs this is not the same operation as <=
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f64 > S1.f64)
# --- end pseudocode ---
def _VOPCOp_V_CMP_NLE_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = !(S0.f64 <= S1.f64);
# // With NAN inputs this is not the same operation as >
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f64 <= S1.f64)
# --- end pseudocode ---
def _VOPCOp_V_CMP_NEQ_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is not equal to the second input. Store the result into VCC
# D0.u64[laneId] = !(S0.f64 == S1.f64);
# // With NAN inputs this is not the same operation as !=
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f64 == S1.f64)
# --- end pseudocode ---
def _VOPCOp_V_CMP_NLT_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is not less than the second input. Store the result into VCC
# D0.u64[laneId] = !(S0.f64 < S1.f64);
# // With NAN inputs this is not the same operation as >=
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = not (S0.f64 < S1.f64)
# --- end pseudocode ---
def _VOPCOp_V_CMP_T_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1. Store the result into VCC or a scalar register.
# D0.u64[laneId] = 1'1U;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = 1
# --- end pseudocode ---
def _VOPCOp_V_CMP_LT_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is less than the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.i16 < S1.i16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i16 < S1.i16
# --- end pseudocode ---
def _VOPCOp_V_CMP_EQ_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.i16 == S1.i16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i16 == S1.i16
# --- end pseudocode ---
def _VOPCOp_V_CMP_LE_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.i16 <= S1.i16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i16 <= S1.i16
# --- end pseudocode ---
def _VOPCOp_V_CMP_GT_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is greater than the second input. Store the result into VCC
# D0.u64[laneId] = S0.i16 > S1.i16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i16 > S1.i16
# --- end pseudocode ---
def _VOPCOp_V_CMP_NE_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is not equal to the second input. Store the result into VCC
# D0.u64[laneId] = S0.i16 <> S1.i16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i16 != S1.i16
# --- end pseudocode ---
def _VOPCOp_V_CMP_GE_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.i16 >= S1.i16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i16 >= S1.i16
# --- end pseudocode ---
def _VOPCOp_V_CMP_LT_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is less than the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.u16 < S1.u16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u16 < S1.u16
# --- end pseudocode ---
def _VOPCOp_V_CMP_EQ_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.u16 == S1.u16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u16 == S1.u16
# --- end pseudocode ---
def _VOPCOp_V_CMP_LE_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.u16 <= S1.u16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u16 <= S1.u16
# --- end pseudocode ---
def _VOPCOp_V_CMP_GT_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is greater than the second input. Store the result into VCC
# D0.u64[laneId] = S0.u16 > S1.u16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u16 > S1.u16
# --- end pseudocode ---
def _VOPCOp_V_CMP_NE_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is not equal to the second input. Store the result into VCC
# D0.u64[laneId] = S0.u16 <> S1.u16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u16 != S1.u16
# --- end pseudocode ---
def _VOPCOp_V_CMP_GE_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.u16 >= S1.u16;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u16 >= S1.u16
# --- end pseudocode ---
def _VOPCOp_V_CMP_F_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 0. Store the result into VCC or a scalar register.
# D0.u64[laneId] = 1'0U;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = 0
# --- end pseudocode ---
def _VOPCOp_V_CMP_LT_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is less than the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.i32 < S1.i32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i32 < S1.i32
# --- end pseudocode ---
def _VOPCOp_V_CMP_EQ_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.i32 == S1.i32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i32 == S1.i32
# --- end pseudocode ---
def _VOPCOp_V_CMP_LE_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.i32 <= S1.i32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i32 <= S1.i32
# --- end pseudocode ---
def _VOPCOp_V_CMP_GT_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is greater than the second input. Store the result into VCC
# D0.u64[laneId] = S0.i32 > S1.i32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i32 > S1.i32
# --- end pseudocode ---
def _VOPCOp_V_CMP_NE_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is not equal to the second input. Store the result into VCC
# D0.u64[laneId] = S0.i32 <> S1.i32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i32 != S1.i32
# --- end pseudocode ---
def _VOPCOp_V_CMP_GE_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.i32 >= S1.i32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i32 >= S1.i32
# --- end pseudocode ---
def _VOPCOp_V_CMP_T_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1. Store the result into VCC or a scalar register.
# D0.u64[laneId] = 1'1U;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = 1
# --- end pseudocode ---
def _VOPCOp_V_CMP_F_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 0. Store the result into VCC or a scalar register.
# D0.u64[laneId] = 1'0U;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = 0
# --- end pseudocode ---
def _VOPCOp_V_CMP_LT_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is less than the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.u32 < S1.u32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u32 < S1.u32
# --- end pseudocode ---
def _VOPCOp_V_CMP_EQ_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.u32 == S1.u32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u32 == S1.u32
# --- end pseudocode ---
def _VOPCOp_V_CMP_LE_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.u32 <= S1.u32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u32 <= S1.u32
# --- end pseudocode ---
def _VOPCOp_V_CMP_GT_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is greater than the second input. Store the result into VCC
# D0.u64[laneId] = S0.u32 > S1.u32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u32 > S1.u32
# --- end pseudocode ---
def _VOPCOp_V_CMP_NE_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is not equal to the second input. Store the result into VCC
# D0.u64[laneId] = S0.u32 <> S1.u32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u32 != S1.u32
# --- end pseudocode ---
def _VOPCOp_V_CMP_GE_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.u32 >= S1.u32;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u32 >= S1.u32
# --- end pseudocode ---
def _VOPCOp_V_CMP_T_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1. Store the result into VCC or a scalar register.
# D0.u64[laneId] = 1'1U;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = 1
# --- end pseudocode ---
def _VOPCOp_V_CMP_F_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 0. Store the result into VCC or a scalar register.
# D0.u64[laneId] = 1'0U;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = 0
# --- end pseudocode ---
def _VOPCOp_V_CMP_LT_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is less than the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.i64 < S1.i64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i64 < S1.i64
# --- end pseudocode ---
def _VOPCOp_V_CMP_EQ_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.i64 == S1.i64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i64 == S1.i64
# --- end pseudocode ---
def _VOPCOp_V_CMP_LE_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.i64 <= S1.i64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i64 <= S1.i64
# --- end pseudocode ---
def _VOPCOp_V_CMP_GT_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is greater than the second input. Store the result into VCC
# D0.u64[laneId] = S0.i64 > S1.i64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i64 > S1.i64
# --- end pseudocode ---
def _VOPCOp_V_CMP_NE_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is not equal to the second input. Store the result into VCC
# D0.u64[laneId] = S0.i64 <> S1.i64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i64 != S1.i64
# --- end pseudocode ---
def _VOPCOp_V_CMP_GE_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.i64 >= S1.i64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.i64 >= S1.i64
# --- end pseudocode ---
def _VOPCOp_V_CMP_T_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1. Store the result into VCC or a scalar register.
# D0.u64[laneId] = 1'1U;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = 1
# --- end pseudocode ---
def _VOPCOp_V_CMP_F_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 0. Store the result into VCC or a scalar register.
# D0.u64[laneId] = 1'0U;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = 0
# --- end pseudocode ---
def _VOPCOp_V_CMP_LT_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is less than the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.u64 < S1.u64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u64 < S1.u64
# --- end pseudocode ---
def _VOPCOp_V_CMP_EQ_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into VCC or a
# D0.u64[laneId] = S0.u64 == S1.u64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u64 == S1.u64
# --- end pseudocode ---
def _VOPCOp_V_CMP_LE_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.u64 <= S1.u64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u64 <= S1.u64
# --- end pseudocode ---
def _VOPCOp_V_CMP_GT_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is greater than the second input. Store the result into VCC
# D0.u64[laneId] = S0.u64 > S1.u64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u64 > S1.u64
# --- end pseudocode ---
def _VOPCOp_V_CMP_NE_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is not equal to the second input. Store the result into VCC
# D0.u64[laneId] = S0.u64 <> S1.u64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u64 != S1.u64
# --- end pseudocode ---
def _VOPCOp_V_CMP_GE_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# D0.u64[laneId] = S0.u64 >= S1.u64;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = S0.u64 >= S1.u64
# --- end pseudocode ---
def _VOPCOp_V_CMP_T_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1. Store the result into VCC or a scalar register.
# D0.u64[laneId] = 1'1U;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
D0.u64[laneId] = 1
# --- end pseudocode ---
def _VOPCOp_V_CMP_CLASS_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# half-precision float, and set the per-lane condition code to the result. Store the result into VCC or a scalar
# S1.u[0] value is a signaling NAN.
# S1.u[1] value is a quiet NAN.
# S1.u[2] value is negative infinity.
# S1.u[3] value is a negative normal value.
# S1.u[4] value is a negative denormal value.
# S1.u[5] value is negative zero.
# S1.u[6] value is positive zero.
# S1.u[7] value is a positive denormal value.
# S1.u[8] value is a positive normal value.
# S1.u[9] value is positive infinity.
# declare result : 1'U;
# if isSignalNAN(64'F(S0.f16)) then
# result = S1.u32[0]
# elsif isQuietNAN(64'F(S0.f16)) then
# result = S1.u32[1]
# elsif exponent(S0.f16) == 31 then
# // +-INF
# result = S1.u32[sign(S0.f16) ? 2 : 9]
# elsif exponent(S0.f16) > 0 then
# // +-normal value
# result = S1.u32[sign(S0.f16) ? 3 : 8]
# elsif 64'F(abs(S0.f16)) > 0.0 then
# // +-denormal value
# result = S1.u32[sign(S0.f16) ? 4 : 7]
# else
# // +-0.0
# result = S1.u32[sign(S0.f16) ? 5 : 6]
# endif;
# D0.u64[laneId] = result;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
if isSignalNAN(F(S0.f16)):
result = S1.u32[0]
elif isQuietNAN(F(S0.f16)):
result = S1.u32[1]
elif exponent(S0.f16) == 31:
result = S1.u32[((2) if (sign(S0.f16)) else (9))]
elif exponent(S0.f16) > 0:
result = S1.u32[((3) if (sign(S0.f16)) else (8))]
elif F(abs(S0.f16)) > 0.0:
result = S1.u32[((4) if (sign(S0.f16)) else (7))]
else:
result = S1.u32[((5) if (sign(S0.f16)) else (6))]
D0.u64[laneId] = result
# --- end pseudocode ---
def _VOPCOp_V_CMP_CLASS_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# single-precision float, and set the per-lane condition code to the result. Store the result into VCC or a scalar
# S1.u[0] value is a signaling NAN.
# S1.u[1] value is a quiet NAN.
# S1.u[2] value is negative infinity.
# S1.u[3] value is a negative normal value.
# S1.u[4] value is a negative denormal value.
# S1.u[5] value is negative zero.
# S1.u[6] value is positive zero.
# S1.u[7] value is a positive denormal value.
# S1.u[8] value is a positive normal value.
# S1.u[9] value is positive infinity.
# declare result : 1'U;
# if isSignalNAN(64'F(S0.f32)) then
# result = S1.u32[0]
# elsif isQuietNAN(64'F(S0.f32)) then
# result = S1.u32[1]
# elsif exponent(S0.f32) == 255 then
# // +-INF
# result = S1.u32[sign(S0.f32) ? 2 : 9]
# elsif exponent(S0.f32) > 0 then
# // +-normal value
# result = S1.u32[sign(S0.f32) ? 3 : 8]
# elsif 64'F(abs(S0.f32)) > 0.0 then
# // +-denormal value
# result = S1.u32[sign(S0.f32) ? 4 : 7]
# else
# // +-0.0
# result = S1.u32[sign(S0.f32) ? 5 : 6]
# endif;
# D0.u64[laneId] = result;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
if isSignalNAN(F(S0.f32)):
result = S1.u32[0]
elif isQuietNAN(F(S0.f32)):
result = S1.u32[1]
elif exponent(S0.f32) == 255:
result = S1.u32[((2) if (sign(S0.f32)) else (9))]
elif exponent(S0.f32) > 0:
result = S1.u32[((3) if (sign(S0.f32)) else (8))]
elif F(abs(S0.f32)) > 0.0:
result = S1.u32[((4) if (sign(S0.f32)) else (7))]
else:
result = S1.u32[((5) if (sign(S0.f32)) else (6))]
D0.u64[laneId] = result
# --- end pseudocode ---
def _VOPCOp_V_CMP_CLASS_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# double-precision float, and set the per-lane condition code to the result. Store the result into VCC or a scalar
# S1.u[0] value is a signaling NAN.
# S1.u[1] value is a quiet NAN.
# S1.u[2] value is negative infinity.
# S1.u[3] value is a negative normal value.
# S1.u[4] value is a negative denormal value.
# S1.u[5] value is negative zero.
# S1.u[6] value is positive zero.
# S1.u[7] value is a positive denormal value.
# S1.u[8] value is a positive normal value.
# S1.u[9] value is positive infinity.
# declare result : 1'U;
# if isSignalNAN(S0.f64) then
# result = S1.u32[0]
# elsif isQuietNAN(S0.f64) then
# result = S1.u32[1]
# elsif exponent(S0.f64) == 2047 then
# // +-INF
# result = S1.u32[sign(S0.f64) ? 2 : 9]
# elsif exponent(S0.f64) > 0 then
# // +-normal value
# result = S1.u32[sign(S0.f64) ? 3 : 8]
# elsif abs(S0.f64) > 0.0 then
# // +-denormal value
# result = S1.u32[sign(S0.f64) ? 4 : 7]
# else
# // +-0.0
# result = S1.u32[sign(S0.f64) ? 5 : 6]
# endif;
# D0.u64[laneId] = result;
# // D0 = VCC in VOPC encoding.
# --- compiled pseudocode ---
if isSignalNAN(S0.f64):
result = S1.u32[0]
elif isQuietNAN(S0.f64):
result = S1.u32[1]
elif exponent(S0.f64) == 2047:
result = S1.u32[((2) if (sign(S0.f64)) else (9))]
elif exponent(S0.f64) > 0:
result = S1.u32[((3) if (sign(S0.f64)) else (8))]
elif abs(S0.f64) > 0.0:
result = S1.u32[((4) if (sign(S0.f64)) else (7))]
else:
result = S1.u32[((5) if (sign(S0.f64)) else (6))]
D0.u64[laneId] = result
# --- end pseudocode ---
def _VOPCOp_V_CMPX_F_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = 1'0U
# --- compiled pseudocode ---
EXEC.u64[laneId] = 0
# --- end pseudocode ---
def _VOPCOp_V_CMPX_LT_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f16 < S1.f16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f16 < S1.f16
# --- end pseudocode ---
def _VOPCOp_V_CMPX_EQ_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into the EXEC
# EXEC.u64[laneId] = S0.f16 == S1.f16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f16 == S1.f16
# --- end pseudocode ---
def _VOPCOp_V_CMPX_LE_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f16 <= S1.f16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f16 <= S1.f16
# --- end pseudocode ---
def _VOPCOp_V_CMPX_GT_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f16 > S1.f16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f16 > S1.f16
# --- end pseudocode ---
def _VOPCOp_V_CMPX_LG_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f16 <> S1.f16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f16 != S1.f16
# --- end pseudocode ---
def _VOPCOp_V_CMPX_GE_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f16 >= S1.f16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f16 >= S1.f16
# --- end pseudocode ---
def _VOPCOp_V_CMPX_O_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = (!isNAN(64'F(S0.f16)) && !isNAN(64'F(S1.f16)))
# --- compiled pseudocode ---
EXEC.u64[laneId] = ( not isNAN(F(S0.f16)) and not isNAN(F(S1.f16)))
# --- end pseudocode ---
def _VOPCOp_V_CMPX_U_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = (isNAN(64'F(S0.f16)) || isNAN(64'F(S1.f16)))
# --- compiled pseudocode ---
EXEC.u64[laneId] = (isNAN(F(S0.f16)) or isNAN(F(S1.f16)))
# --- end pseudocode ---
def _VOPCOp_V_CMPX_NGE_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f16 >= S1.f16);
# // With NAN inputs this is not the same operation as <
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f16 >= S1.f16)
# --- end pseudocode ---
def _VOPCOp_V_CMPX_NLG_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f16 <> S1.f16);
# // With NAN inputs this is not the same operation as ==
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f16 != S1.f16)
# --- end pseudocode ---
def _VOPCOp_V_CMPX_NGT_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f16 > S1.f16);
# // With NAN inputs this is not the same operation as <=
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f16 > S1.f16)
# --- end pseudocode ---
def _VOPCOp_V_CMPX_NLE_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f16 <= S1.f16);
# // With NAN inputs this is not the same operation as >
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f16 <= S1.f16)
# --- end pseudocode ---
def _VOPCOp_V_CMPX_NEQ_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f16 == S1.f16);
# // With NAN inputs this is not the same operation as !=
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f16 == S1.f16)
# --- end pseudocode ---
def _VOPCOp_V_CMPX_NLT_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f16 < S1.f16);
# // With NAN inputs this is not the same operation as >=
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f16 < S1.f16)
# --- end pseudocode ---
def _VOPCOp_V_CMPX_T_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = 1'1U
# --- compiled pseudocode ---
EXEC.u64[laneId] = 1
# --- end pseudocode ---
def _VOPCOp_V_CMPX_F_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = 1'0U
# --- compiled pseudocode ---
EXEC.u64[laneId] = 0
# --- end pseudocode ---
def _VOPCOp_V_CMPX_LT_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f32 < S1.f32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f32 < S1.f32
# --- end pseudocode ---
def _VOPCOp_V_CMPX_EQ_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into the EXEC
# EXEC.u64[laneId] = S0.f32 == S1.f32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f32 == S1.f32
# --- end pseudocode ---
def _VOPCOp_V_CMPX_LE_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f32 <= S1.f32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f32 <= S1.f32
# --- end pseudocode ---
def _VOPCOp_V_CMPX_GT_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f32 > S1.f32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f32 > S1.f32
# --- end pseudocode ---
def _VOPCOp_V_CMPX_LG_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f32 <> S1.f32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f32 != S1.f32
# --- end pseudocode ---
def _VOPCOp_V_CMPX_GE_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f32 >= S1.f32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f32 >= S1.f32
# --- end pseudocode ---
def _VOPCOp_V_CMPX_O_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = (!isNAN(64'F(S0.f32)) && !isNAN(64'F(S1.f32)))
# --- compiled pseudocode ---
EXEC.u64[laneId] = ( not isNAN(F(S0.f32)) and not isNAN(F(S1.f32)))
# --- end pseudocode ---
def _VOPCOp_V_CMPX_U_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = (isNAN(64'F(S0.f32)) || isNAN(64'F(S1.f32)))
# --- compiled pseudocode ---
EXEC.u64[laneId] = (isNAN(F(S0.f32)) or isNAN(F(S1.f32)))
# --- end pseudocode ---
def _VOPCOp_V_CMPX_NGE_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f32 >= S1.f32);
# // With NAN inputs this is not the same operation as <
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f32 >= S1.f32)
# --- end pseudocode ---
def _VOPCOp_V_CMPX_NLG_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f32 <> S1.f32);
# // With NAN inputs this is not the same operation as ==
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f32 != S1.f32)
# --- end pseudocode ---
def _VOPCOp_V_CMPX_NGT_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f32 > S1.f32);
# // With NAN inputs this is not the same operation as <=
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f32 > S1.f32)
# --- end pseudocode ---
def _VOPCOp_V_CMPX_NLE_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f32 <= S1.f32);
# // With NAN inputs this is not the same operation as >
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f32 <= S1.f32)
# --- end pseudocode ---
def _VOPCOp_V_CMPX_NEQ_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f32 == S1.f32);
# // With NAN inputs this is not the same operation as !=
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f32 == S1.f32)
# --- end pseudocode ---
def _VOPCOp_V_CMPX_NLT_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f32 < S1.f32);
# // With NAN inputs this is not the same operation as >=
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f32 < S1.f32)
# --- end pseudocode ---
def _VOPCOp_V_CMPX_T_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = 1'1U
# --- compiled pseudocode ---
EXEC.u64[laneId] = 1
# --- end pseudocode ---
def _VOPCOp_V_CMPX_F_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = 1'0U
# --- compiled pseudocode ---
EXEC.u64[laneId] = 0
# --- end pseudocode ---
def _VOPCOp_V_CMPX_LT_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f64 < S1.f64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f64 < S1.f64
# --- end pseudocode ---
def _VOPCOp_V_CMPX_EQ_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into the EXEC
# EXEC.u64[laneId] = S0.f64 == S1.f64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f64 == S1.f64
# --- end pseudocode ---
def _VOPCOp_V_CMPX_LE_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f64 <= S1.f64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f64 <= S1.f64
# --- end pseudocode ---
def _VOPCOp_V_CMPX_GT_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f64 > S1.f64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f64 > S1.f64
# --- end pseudocode ---
def _VOPCOp_V_CMPX_LG_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f64 <> S1.f64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f64 != S1.f64
# --- end pseudocode ---
def _VOPCOp_V_CMPX_GE_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.f64 >= S1.f64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.f64 >= S1.f64
# --- end pseudocode ---
def _VOPCOp_V_CMPX_O_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = (!isNAN(S0.f64) && !isNAN(S1.f64))
# --- compiled pseudocode ---
EXEC.u64[laneId] = ( not isNAN(S0.f64) and not isNAN(S1.f64))
# --- end pseudocode ---
def _VOPCOp_V_CMPX_U_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = (isNAN(S0.f64) || isNAN(S1.f64))
# --- compiled pseudocode ---
EXEC.u64[laneId] = (isNAN(S0.f64) or isNAN(S1.f64))
# --- end pseudocode ---
def _VOPCOp_V_CMPX_NGE_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f64 >= S1.f64);
# // With NAN inputs this is not the same operation as <
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f64 >= S1.f64)
# --- end pseudocode ---
def _VOPCOp_V_CMPX_NLG_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f64 <> S1.f64);
# // With NAN inputs this is not the same operation as ==
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f64 != S1.f64)
# --- end pseudocode ---
def _VOPCOp_V_CMPX_NGT_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f64 > S1.f64);
# // With NAN inputs this is not the same operation as <=
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f64 > S1.f64)
# --- end pseudocode ---
def _VOPCOp_V_CMPX_NLE_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f64 <= S1.f64);
# // With NAN inputs this is not the same operation as >
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f64 <= S1.f64)
# --- end pseudocode ---
def _VOPCOp_V_CMPX_NEQ_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f64 == S1.f64);
# // With NAN inputs this is not the same operation as !=
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f64 == S1.f64)
# --- end pseudocode ---
def _VOPCOp_V_CMPX_NLT_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = !(S0.f64 < S1.f64);
# // With NAN inputs this is not the same operation as >=
# --- compiled pseudocode ---
EXEC.u64[laneId] = not (S0.f64 < S1.f64)
# --- end pseudocode ---
def _VOPCOp_V_CMPX_T_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = 1'1U
# --- compiled pseudocode ---
EXEC.u64[laneId] = 1
# --- end pseudocode ---
def _VOPCOp_V_CMPX_LT_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i16 < S1.i16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i16 < S1.i16
# --- end pseudocode ---
def _VOPCOp_V_CMPX_EQ_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into the EXEC
# EXEC.u64[laneId] = S0.i16 == S1.i16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i16 == S1.i16
# --- end pseudocode ---
def _VOPCOp_V_CMPX_LE_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i16 <= S1.i16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i16 <= S1.i16
# --- end pseudocode ---
def _VOPCOp_V_CMPX_GT_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i16 > S1.i16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i16 > S1.i16
# --- end pseudocode ---
def _VOPCOp_V_CMPX_NE_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i16 <> S1.i16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i16 != S1.i16
# --- end pseudocode ---
def _VOPCOp_V_CMPX_GE_I16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i16 >= S1.i16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i16 >= S1.i16
# --- end pseudocode ---
def _VOPCOp_V_CMPX_LT_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u16 < S1.u16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u16 < S1.u16
# --- end pseudocode ---
def _VOPCOp_V_CMPX_EQ_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into the EXEC
# EXEC.u64[laneId] = S0.u16 == S1.u16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u16 == S1.u16
# --- end pseudocode ---
def _VOPCOp_V_CMPX_LE_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u16 <= S1.u16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u16 <= S1.u16
# --- end pseudocode ---
def _VOPCOp_V_CMPX_GT_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u16 > S1.u16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u16 > S1.u16
# --- end pseudocode ---
def _VOPCOp_V_CMPX_NE_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u16 <> S1.u16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u16 != S1.u16
# --- end pseudocode ---
def _VOPCOp_V_CMPX_GE_U16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u16 >= S1.u16
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u16 >= S1.u16
# --- end pseudocode ---
def _VOPCOp_V_CMPX_F_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = 1'0U
# --- compiled pseudocode ---
EXEC.u64[laneId] = 0
# --- end pseudocode ---
def _VOPCOp_V_CMPX_LT_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i32 < S1.i32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i32 < S1.i32
# --- end pseudocode ---
def _VOPCOp_V_CMPX_EQ_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into the EXEC
# EXEC.u64[laneId] = S0.i32 == S1.i32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i32 == S1.i32
# --- end pseudocode ---
def _VOPCOp_V_CMPX_LE_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i32 <= S1.i32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i32 <= S1.i32
# --- end pseudocode ---
def _VOPCOp_V_CMPX_GT_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i32 > S1.i32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i32 > S1.i32
# --- end pseudocode ---
def _VOPCOp_V_CMPX_NE_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i32 <> S1.i32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i32 != S1.i32
# --- end pseudocode ---
def _VOPCOp_V_CMPX_GE_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i32 >= S1.i32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i32 >= S1.i32
# --- end pseudocode ---
def _VOPCOp_V_CMPX_T_I32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = 1'1U
# --- compiled pseudocode ---
EXEC.u64[laneId] = 1
# --- end pseudocode ---
def _VOPCOp_V_CMPX_F_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = 1'0U
# --- compiled pseudocode ---
EXEC.u64[laneId] = 0
# --- end pseudocode ---
def _VOPCOp_V_CMPX_LT_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u32 < S1.u32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u32 < S1.u32
# --- end pseudocode ---
def _VOPCOp_V_CMPX_EQ_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into the EXEC
# EXEC.u64[laneId] = S0.u32 == S1.u32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u32 == S1.u32
# --- end pseudocode ---
def _VOPCOp_V_CMPX_LE_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u32 <= S1.u32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u32 <= S1.u32
# --- end pseudocode ---
def _VOPCOp_V_CMPX_GT_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u32 > S1.u32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u32 > S1.u32
# --- end pseudocode ---
def _VOPCOp_V_CMPX_NE_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u32 <> S1.u32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u32 != S1.u32
# --- end pseudocode ---
def _VOPCOp_V_CMPX_GE_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u32 >= S1.u32
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u32 >= S1.u32
# --- end pseudocode ---
def _VOPCOp_V_CMPX_T_U32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = 1'1U
# --- compiled pseudocode ---
EXEC.u64[laneId] = 1
# --- end pseudocode ---
def _VOPCOp_V_CMPX_F_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = 1'0U
# --- compiled pseudocode ---
EXEC.u64[laneId] = 0
# --- end pseudocode ---
def _VOPCOp_V_CMPX_LT_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i64 < S1.i64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i64 < S1.i64
# --- end pseudocode ---
def _VOPCOp_V_CMPX_EQ_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into the EXEC
# EXEC.u64[laneId] = S0.i64 == S1.i64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i64 == S1.i64
# --- end pseudocode ---
def _VOPCOp_V_CMPX_LE_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i64 <= S1.i64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i64 <= S1.i64
# --- end pseudocode ---
def _VOPCOp_V_CMPX_GT_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i64 > S1.i64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i64 > S1.i64
# --- end pseudocode ---
def _VOPCOp_V_CMPX_NE_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i64 <> S1.i64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i64 != S1.i64
# --- end pseudocode ---
def _VOPCOp_V_CMPX_GE_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.i64 >= S1.i64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.i64 >= S1.i64
# --- end pseudocode ---
def _VOPCOp_V_CMPX_T_I64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = 1'1U
# --- compiled pseudocode ---
EXEC.u64[laneId] = 1
# --- end pseudocode ---
def _VOPCOp_V_CMPX_F_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = 1'0U
# --- compiled pseudocode ---
EXEC.u64[laneId] = 0
# --- end pseudocode ---
def _VOPCOp_V_CMPX_LT_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u64 < S1.u64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u64 < S1.u64
# --- end pseudocode ---
def _VOPCOp_V_CMPX_EQ_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# Set the per-lane condition code to 1 iff the first input is equal to the second input. Store the result into the EXEC
# EXEC.u64[laneId] = S0.u64 == S1.u64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u64 == S1.u64
# --- end pseudocode ---
def _VOPCOp_V_CMPX_LE_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u64 <= S1.u64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u64 <= S1.u64
# --- end pseudocode ---
def _VOPCOp_V_CMPX_GT_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u64 > S1.u64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u64 > S1.u64
# --- end pseudocode ---
def _VOPCOp_V_CMPX_NE_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u64 <> S1.u64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u64 != S1.u64
# --- end pseudocode ---
def _VOPCOp_V_CMPX_GE_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = S0.u64 >= S1.u64
# --- compiled pseudocode ---
EXEC.u64[laneId] = S0.u64 >= S1.u64
# --- end pseudocode ---
def _VOPCOp_V_CMPX_T_U64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# EXEC.u64[laneId] = 1'1U
# --- compiled pseudocode ---
EXEC.u64[laneId] = 1
# --- end pseudocode ---
def _VOPCOp_V_CMPX_CLASS_F16(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# S1.u[0] value is a signaling NAN.
# S1.u[1] value is a quiet NAN.
# S1.u[2] value is negative infinity.
# S1.u[3] value is a negative normal value.
# S1.u[4] value is a negative denormal value.
# S1.u[5] value is negative zero.
# S1.u[6] value is positive zero.
# S1.u[7] value is a positive denormal value.
# S1.u[8] value is a positive normal value.
# S1.u[9] value is positive infinity.
# declare result : 1'U;
# if isSignalNAN(64'F(S0.f16)) then
# result = S1.u32[0]
# elsif isQuietNAN(64'F(S0.f16)) then
# result = S1.u32[1]
# elsif exponent(S0.f16) == 31 then
# // +-INF
# result = S1.u32[sign(S0.f16) ? 2 : 9]
# elsif exponent(S0.f16) > 0 then
# // +-normal value
# result = S1.u32[sign(S0.f16) ? 3 : 8]
# elsif 64'F(abs(S0.f16)) > 0.0 then
# // +-denormal value
# result = S1.u32[sign(S0.f16) ? 4 : 7]
# else
# // +-0.0
# result = S1.u32[sign(S0.f16) ? 5 : 6]
# endif;
# EXEC.u64[laneId] = result
# --- compiled pseudocode ---
if isSignalNAN(F(S0.f16)):
result = S1.u32[0]
elif isQuietNAN(F(S0.f16)):
result = S1.u32[1]
elif exponent(S0.f16) == 31:
result = S1.u32[((2) if (sign(S0.f16)) else (9))]
elif exponent(S0.f16) > 0:
result = S1.u32[((3) if (sign(S0.f16)) else (8))]
elif F(abs(S0.f16)) > 0.0:
result = S1.u32[((4) if (sign(S0.f16)) else (7))]
else:
result = S1.u32[((5) if (sign(S0.f16)) else (6))]
EXEC.u64[laneId] = result
# --- end pseudocode ---
def _VOPCOp_V_CMPX_CLASS_F32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# S1.u[0] value is a signaling NAN.
# S1.u[1] value is a quiet NAN.
# S1.u[2] value is negative infinity.
# S1.u[3] value is a negative normal value.
# S1.u[4] value is a negative denormal value.
# S1.u[5] value is negative zero.
# S1.u[6] value is positive zero.
# S1.u[7] value is a positive denormal value.
# S1.u[8] value is a positive normal value.
# S1.u[9] value is positive infinity.
# declare result : 1'U;
# if isSignalNAN(64'F(S0.f32)) then
# result = S1.u32[0]
# elsif isQuietNAN(64'F(S0.f32)) then
# result = S1.u32[1]
# elsif exponent(S0.f32) == 255 then
# // +-INF
# result = S1.u32[sign(S0.f32) ? 2 : 9]
# elsif exponent(S0.f32) > 0 then
# // +-normal value
# result = S1.u32[sign(S0.f32) ? 3 : 8]
# elsif 64'F(abs(S0.f32)) > 0.0 then
# // +-denormal value
# result = S1.u32[sign(S0.f32) ? 4 : 7]
# else
# // +-0.0
# result = S1.u32[sign(S0.f32) ? 5 : 6]
# endif;
# EXEC.u64[laneId] = result
# --- compiled pseudocode ---
if isSignalNAN(F(S0.f32)):
result = S1.u32[0]
elif isQuietNAN(F(S0.f32)):
result = S1.u32[1]
elif exponent(S0.f32) == 255:
result = S1.u32[((2) if (sign(S0.f32)) else (9))]
elif exponent(S0.f32) > 0:
result = S1.u32[((3) if (sign(S0.f32)) else (8))]
elif F(abs(S0.f32)) > 0.0:
result = S1.u32[((4) if (sign(S0.f32)) else (7))]
else:
result = S1.u32[((5) if (sign(S0.f32)) else (6))]
EXEC.u64[laneId] = result
# --- end pseudocode ---
def _VOPCOp_V_CMPX_CLASS_F64(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
# S1.u[0] value is a signaling NAN.
# S1.u[1] value is a quiet NAN.
# S1.u[2] value is negative infinity.
# S1.u[3] value is a negative normal value.
# S1.u[4] value is a negative denormal value.
# S1.u[5] value is negative zero.
# S1.u[6] value is positive zero.
# S1.u[7] value is a positive denormal value.
# S1.u[8] value is a positive normal value.
# S1.u[9] value is positive infinity.
# declare result : 1'U;
# if isSignalNAN(S0.f64) then
# result = S1.u32[0]
# elsif isQuietNAN(S0.f64) then
# result = S1.u32[1]
# elsif exponent(S0.f64) == 2047 then
# // +-INF
# result = S1.u32[sign(S0.f64) ? 2 : 9]
# elsif exponent(S0.f64) > 0 then
# // +-normal value
# result = S1.u32[sign(S0.f64) ? 3 : 8]
# elsif abs(S0.f64) > 0.0 then
# // +-denormal value
# result = S1.u32[sign(S0.f64) ? 4 : 7]
# else
# // +-0.0
# result = S1.u32[sign(S0.f64) ? 5 : 6]
# endif;
# EXEC.u64[laneId] = result
# --- compiled pseudocode ---
if isSignalNAN(S0.f64):
result = S1.u32[0]
elif isQuietNAN(S0.f64):
result = S1.u32[1]
elif exponent(S0.f64) == 2047:
result = S1.u32[((2) if (sign(S0.f64)) else (9))]
elif exponent(S0.f64) > 0:
result = S1.u32[((3) if (sign(S0.f64)) else (8))]
elif abs(S0.f64) > 0.0:
result = S1.u32[((4) if (sign(S0.f64)) else (7))]
else:
result = S1.u32[((5) if (sign(S0.f64)) else (6))]
EXEC.u64[laneId] = result
# --- end pseudocode ---
VOPCOp_FUNCTIONS = {
VOPCOp.V_CMP_F_F16: _VOPCOp_V_CMP_F_F16,
VOPCOp.V_CMP_LT_F16: _VOPCOp_V_CMP_LT_F16,
VOPCOp.V_CMP_EQ_F16: _VOPCOp_V_CMP_EQ_F16,
VOPCOp.V_CMP_LE_F16: _VOPCOp_V_CMP_LE_F16,
VOPCOp.V_CMP_GT_F16: _VOPCOp_V_CMP_GT_F16,
VOPCOp.V_CMP_LG_F16: _VOPCOp_V_CMP_LG_F16,
VOPCOp.V_CMP_GE_F16: _VOPCOp_V_CMP_GE_F16,
VOPCOp.V_CMP_O_F16: _VOPCOp_V_CMP_O_F16,
VOPCOp.V_CMP_U_F16: _VOPCOp_V_CMP_U_F16,
VOPCOp.V_CMP_NGE_F16: _VOPCOp_V_CMP_NGE_F16,
VOPCOp.V_CMP_NLG_F16: _VOPCOp_V_CMP_NLG_F16,
VOPCOp.V_CMP_NGT_F16: _VOPCOp_V_CMP_NGT_F16,
VOPCOp.V_CMP_NLE_F16: _VOPCOp_V_CMP_NLE_F16,
VOPCOp.V_CMP_NEQ_F16: _VOPCOp_V_CMP_NEQ_F16,
VOPCOp.V_CMP_NLT_F16: _VOPCOp_V_CMP_NLT_F16,
VOPCOp.V_CMP_T_F16: _VOPCOp_V_CMP_T_F16,
VOPCOp.V_CMP_F_F32: _VOPCOp_V_CMP_F_F32,
VOPCOp.V_CMP_LT_F32: _VOPCOp_V_CMP_LT_F32,
VOPCOp.V_CMP_EQ_F32: _VOPCOp_V_CMP_EQ_F32,
VOPCOp.V_CMP_LE_F32: _VOPCOp_V_CMP_LE_F32,
VOPCOp.V_CMP_GT_F32: _VOPCOp_V_CMP_GT_F32,
VOPCOp.V_CMP_LG_F32: _VOPCOp_V_CMP_LG_F32,
VOPCOp.V_CMP_GE_F32: _VOPCOp_V_CMP_GE_F32,
VOPCOp.V_CMP_O_F32: _VOPCOp_V_CMP_O_F32,
VOPCOp.V_CMP_U_F32: _VOPCOp_V_CMP_U_F32,
VOPCOp.V_CMP_NGE_F32: _VOPCOp_V_CMP_NGE_F32,
VOPCOp.V_CMP_NLG_F32: _VOPCOp_V_CMP_NLG_F32,
VOPCOp.V_CMP_NGT_F32: _VOPCOp_V_CMP_NGT_F32,
VOPCOp.V_CMP_NLE_F32: _VOPCOp_V_CMP_NLE_F32,
VOPCOp.V_CMP_NEQ_F32: _VOPCOp_V_CMP_NEQ_F32,
VOPCOp.V_CMP_NLT_F32: _VOPCOp_V_CMP_NLT_F32,
VOPCOp.V_CMP_T_F32: _VOPCOp_V_CMP_T_F32,
VOPCOp.V_CMP_F_F64: _VOPCOp_V_CMP_F_F64,
VOPCOp.V_CMP_LT_F64: _VOPCOp_V_CMP_LT_F64,
VOPCOp.V_CMP_EQ_F64: _VOPCOp_V_CMP_EQ_F64,
VOPCOp.V_CMP_LE_F64: _VOPCOp_V_CMP_LE_F64,
VOPCOp.V_CMP_GT_F64: _VOPCOp_V_CMP_GT_F64,
VOPCOp.V_CMP_LG_F64: _VOPCOp_V_CMP_LG_F64,
VOPCOp.V_CMP_GE_F64: _VOPCOp_V_CMP_GE_F64,
VOPCOp.V_CMP_O_F64: _VOPCOp_V_CMP_O_F64,
VOPCOp.V_CMP_U_F64: _VOPCOp_V_CMP_U_F64,
VOPCOp.V_CMP_NGE_F64: _VOPCOp_V_CMP_NGE_F64,
VOPCOp.V_CMP_NLG_F64: _VOPCOp_V_CMP_NLG_F64,
VOPCOp.V_CMP_NGT_F64: _VOPCOp_V_CMP_NGT_F64,
VOPCOp.V_CMP_NLE_F64: _VOPCOp_V_CMP_NLE_F64,
VOPCOp.V_CMP_NEQ_F64: _VOPCOp_V_CMP_NEQ_F64,
VOPCOp.V_CMP_NLT_F64: _VOPCOp_V_CMP_NLT_F64,
VOPCOp.V_CMP_T_F64: _VOPCOp_V_CMP_T_F64,
VOPCOp.V_CMP_LT_I16: _VOPCOp_V_CMP_LT_I16,
VOPCOp.V_CMP_EQ_I16: _VOPCOp_V_CMP_EQ_I16,
VOPCOp.V_CMP_LE_I16: _VOPCOp_V_CMP_LE_I16,
VOPCOp.V_CMP_GT_I16: _VOPCOp_V_CMP_GT_I16,
VOPCOp.V_CMP_NE_I16: _VOPCOp_V_CMP_NE_I16,
VOPCOp.V_CMP_GE_I16: _VOPCOp_V_CMP_GE_I16,
VOPCOp.V_CMP_LT_U16: _VOPCOp_V_CMP_LT_U16,
VOPCOp.V_CMP_EQ_U16: _VOPCOp_V_CMP_EQ_U16,
VOPCOp.V_CMP_LE_U16: _VOPCOp_V_CMP_LE_U16,
VOPCOp.V_CMP_GT_U16: _VOPCOp_V_CMP_GT_U16,
VOPCOp.V_CMP_NE_U16: _VOPCOp_V_CMP_NE_U16,
VOPCOp.V_CMP_GE_U16: _VOPCOp_V_CMP_GE_U16,
VOPCOp.V_CMP_F_I32: _VOPCOp_V_CMP_F_I32,
VOPCOp.V_CMP_LT_I32: _VOPCOp_V_CMP_LT_I32,
VOPCOp.V_CMP_EQ_I32: _VOPCOp_V_CMP_EQ_I32,
VOPCOp.V_CMP_LE_I32: _VOPCOp_V_CMP_LE_I32,
VOPCOp.V_CMP_GT_I32: _VOPCOp_V_CMP_GT_I32,
VOPCOp.V_CMP_NE_I32: _VOPCOp_V_CMP_NE_I32,
VOPCOp.V_CMP_GE_I32: _VOPCOp_V_CMP_GE_I32,
VOPCOp.V_CMP_T_I32: _VOPCOp_V_CMP_T_I32,
VOPCOp.V_CMP_F_U32: _VOPCOp_V_CMP_F_U32,
VOPCOp.V_CMP_LT_U32: _VOPCOp_V_CMP_LT_U32,
VOPCOp.V_CMP_EQ_U32: _VOPCOp_V_CMP_EQ_U32,
VOPCOp.V_CMP_LE_U32: _VOPCOp_V_CMP_LE_U32,
VOPCOp.V_CMP_GT_U32: _VOPCOp_V_CMP_GT_U32,
VOPCOp.V_CMP_NE_U32: _VOPCOp_V_CMP_NE_U32,
VOPCOp.V_CMP_GE_U32: _VOPCOp_V_CMP_GE_U32,
VOPCOp.V_CMP_T_U32: _VOPCOp_V_CMP_T_U32,
VOPCOp.V_CMP_F_I64: _VOPCOp_V_CMP_F_I64,
VOPCOp.V_CMP_LT_I64: _VOPCOp_V_CMP_LT_I64,
VOPCOp.V_CMP_EQ_I64: _VOPCOp_V_CMP_EQ_I64,
VOPCOp.V_CMP_LE_I64: _VOPCOp_V_CMP_LE_I64,
VOPCOp.V_CMP_GT_I64: _VOPCOp_V_CMP_GT_I64,
VOPCOp.V_CMP_NE_I64: _VOPCOp_V_CMP_NE_I64,
VOPCOp.V_CMP_GE_I64: _VOPCOp_V_CMP_GE_I64,
VOPCOp.V_CMP_T_I64: _VOPCOp_V_CMP_T_I64,
VOPCOp.V_CMP_F_U64: _VOPCOp_V_CMP_F_U64,
VOPCOp.V_CMP_LT_U64: _VOPCOp_V_CMP_LT_U64,
VOPCOp.V_CMP_EQ_U64: _VOPCOp_V_CMP_EQ_U64,
VOPCOp.V_CMP_LE_U64: _VOPCOp_V_CMP_LE_U64,
VOPCOp.V_CMP_GT_U64: _VOPCOp_V_CMP_GT_U64,
VOPCOp.V_CMP_NE_U64: _VOPCOp_V_CMP_NE_U64,
VOPCOp.V_CMP_GE_U64: _VOPCOp_V_CMP_GE_U64,
VOPCOp.V_CMP_T_U64: _VOPCOp_V_CMP_T_U64,
VOPCOp.V_CMP_CLASS_F16: _VOPCOp_V_CMP_CLASS_F16,
VOPCOp.V_CMP_CLASS_F32: _VOPCOp_V_CMP_CLASS_F32,
VOPCOp.V_CMP_CLASS_F64: _VOPCOp_V_CMP_CLASS_F64,
VOPCOp.V_CMPX_F_F16: _VOPCOp_V_CMPX_F_F16,
VOPCOp.V_CMPX_LT_F16: _VOPCOp_V_CMPX_LT_F16,
VOPCOp.V_CMPX_EQ_F16: _VOPCOp_V_CMPX_EQ_F16,
VOPCOp.V_CMPX_LE_F16: _VOPCOp_V_CMPX_LE_F16,
VOPCOp.V_CMPX_GT_F16: _VOPCOp_V_CMPX_GT_F16,
VOPCOp.V_CMPX_LG_F16: _VOPCOp_V_CMPX_LG_F16,
VOPCOp.V_CMPX_GE_F16: _VOPCOp_V_CMPX_GE_F16,
VOPCOp.V_CMPX_O_F16: _VOPCOp_V_CMPX_O_F16,
VOPCOp.V_CMPX_U_F16: _VOPCOp_V_CMPX_U_F16,
VOPCOp.V_CMPX_NGE_F16: _VOPCOp_V_CMPX_NGE_F16,
VOPCOp.V_CMPX_NLG_F16: _VOPCOp_V_CMPX_NLG_F16,
VOPCOp.V_CMPX_NGT_F16: _VOPCOp_V_CMPX_NGT_F16,
VOPCOp.V_CMPX_NLE_F16: _VOPCOp_V_CMPX_NLE_F16,
VOPCOp.V_CMPX_NEQ_F16: _VOPCOp_V_CMPX_NEQ_F16,
VOPCOp.V_CMPX_NLT_F16: _VOPCOp_V_CMPX_NLT_F16,
VOPCOp.V_CMPX_T_F16: _VOPCOp_V_CMPX_T_F16,
VOPCOp.V_CMPX_F_F32: _VOPCOp_V_CMPX_F_F32,
VOPCOp.V_CMPX_LT_F32: _VOPCOp_V_CMPX_LT_F32,
VOPCOp.V_CMPX_EQ_F32: _VOPCOp_V_CMPX_EQ_F32,
VOPCOp.V_CMPX_LE_F32: _VOPCOp_V_CMPX_LE_F32,
VOPCOp.V_CMPX_GT_F32: _VOPCOp_V_CMPX_GT_F32,
VOPCOp.V_CMPX_LG_F32: _VOPCOp_V_CMPX_LG_F32,
VOPCOp.V_CMPX_GE_F32: _VOPCOp_V_CMPX_GE_F32,
VOPCOp.V_CMPX_O_F32: _VOPCOp_V_CMPX_O_F32,
VOPCOp.V_CMPX_U_F32: _VOPCOp_V_CMPX_U_F32,
VOPCOp.V_CMPX_NGE_F32: _VOPCOp_V_CMPX_NGE_F32,
VOPCOp.V_CMPX_NLG_F32: _VOPCOp_V_CMPX_NLG_F32,
VOPCOp.V_CMPX_NGT_F32: _VOPCOp_V_CMPX_NGT_F32,
VOPCOp.V_CMPX_NLE_F32: _VOPCOp_V_CMPX_NLE_F32,
VOPCOp.V_CMPX_NEQ_F32: _VOPCOp_V_CMPX_NEQ_F32,
VOPCOp.V_CMPX_NLT_F32: _VOPCOp_V_CMPX_NLT_F32,
VOPCOp.V_CMPX_T_F32: _VOPCOp_V_CMPX_T_F32,
VOPCOp.V_CMPX_F_F64: _VOPCOp_V_CMPX_F_F64,
VOPCOp.V_CMPX_LT_F64: _VOPCOp_V_CMPX_LT_F64,
VOPCOp.V_CMPX_EQ_F64: _VOPCOp_V_CMPX_EQ_F64,
VOPCOp.V_CMPX_LE_F64: _VOPCOp_V_CMPX_LE_F64,
VOPCOp.V_CMPX_GT_F64: _VOPCOp_V_CMPX_GT_F64,
VOPCOp.V_CMPX_LG_F64: _VOPCOp_V_CMPX_LG_F64,
VOPCOp.V_CMPX_GE_F64: _VOPCOp_V_CMPX_GE_F64,
VOPCOp.V_CMPX_O_F64: _VOPCOp_V_CMPX_O_F64,
VOPCOp.V_CMPX_U_F64: _VOPCOp_V_CMPX_U_F64,
VOPCOp.V_CMPX_NGE_F64: _VOPCOp_V_CMPX_NGE_F64,
VOPCOp.V_CMPX_NLG_F64: _VOPCOp_V_CMPX_NLG_F64,
VOPCOp.V_CMPX_NGT_F64: _VOPCOp_V_CMPX_NGT_F64,
VOPCOp.V_CMPX_NLE_F64: _VOPCOp_V_CMPX_NLE_F64,
VOPCOp.V_CMPX_NEQ_F64: _VOPCOp_V_CMPX_NEQ_F64,
VOPCOp.V_CMPX_NLT_F64: _VOPCOp_V_CMPX_NLT_F64,
VOPCOp.V_CMPX_T_F64: _VOPCOp_V_CMPX_T_F64,
VOPCOp.V_CMPX_LT_I16: _VOPCOp_V_CMPX_LT_I16,
VOPCOp.V_CMPX_EQ_I16: _VOPCOp_V_CMPX_EQ_I16,
VOPCOp.V_CMPX_LE_I16: _VOPCOp_V_CMPX_LE_I16,
VOPCOp.V_CMPX_GT_I16: _VOPCOp_V_CMPX_GT_I16,
VOPCOp.V_CMPX_NE_I16: _VOPCOp_V_CMPX_NE_I16,
VOPCOp.V_CMPX_GE_I16: _VOPCOp_V_CMPX_GE_I16,
VOPCOp.V_CMPX_LT_U16: _VOPCOp_V_CMPX_LT_U16,
VOPCOp.V_CMPX_EQ_U16: _VOPCOp_V_CMPX_EQ_U16,
VOPCOp.V_CMPX_LE_U16: _VOPCOp_V_CMPX_LE_U16,
VOPCOp.V_CMPX_GT_U16: _VOPCOp_V_CMPX_GT_U16,
VOPCOp.V_CMPX_NE_U16: _VOPCOp_V_CMPX_NE_U16,
VOPCOp.V_CMPX_GE_U16: _VOPCOp_V_CMPX_GE_U16,
VOPCOp.V_CMPX_F_I32: _VOPCOp_V_CMPX_F_I32,
VOPCOp.V_CMPX_LT_I32: _VOPCOp_V_CMPX_LT_I32,
VOPCOp.V_CMPX_EQ_I32: _VOPCOp_V_CMPX_EQ_I32,
VOPCOp.V_CMPX_LE_I32: _VOPCOp_V_CMPX_LE_I32,
VOPCOp.V_CMPX_GT_I32: _VOPCOp_V_CMPX_GT_I32,
VOPCOp.V_CMPX_NE_I32: _VOPCOp_V_CMPX_NE_I32,
VOPCOp.V_CMPX_GE_I32: _VOPCOp_V_CMPX_GE_I32,
VOPCOp.V_CMPX_T_I32: _VOPCOp_V_CMPX_T_I32,
VOPCOp.V_CMPX_F_U32: _VOPCOp_V_CMPX_F_U32,
VOPCOp.V_CMPX_LT_U32: _VOPCOp_V_CMPX_LT_U32,
VOPCOp.V_CMPX_EQ_U32: _VOPCOp_V_CMPX_EQ_U32,
VOPCOp.V_CMPX_LE_U32: _VOPCOp_V_CMPX_LE_U32,
VOPCOp.V_CMPX_GT_U32: _VOPCOp_V_CMPX_GT_U32,
VOPCOp.V_CMPX_NE_U32: _VOPCOp_V_CMPX_NE_U32,
VOPCOp.V_CMPX_GE_U32: _VOPCOp_V_CMPX_GE_U32,
VOPCOp.V_CMPX_T_U32: _VOPCOp_V_CMPX_T_U32,
VOPCOp.V_CMPX_F_I64: _VOPCOp_V_CMPX_F_I64,
VOPCOp.V_CMPX_LT_I64: _VOPCOp_V_CMPX_LT_I64,
VOPCOp.V_CMPX_EQ_I64: _VOPCOp_V_CMPX_EQ_I64,
VOPCOp.V_CMPX_LE_I64: _VOPCOp_V_CMPX_LE_I64,
VOPCOp.V_CMPX_GT_I64: _VOPCOp_V_CMPX_GT_I64,
VOPCOp.V_CMPX_NE_I64: _VOPCOp_V_CMPX_NE_I64,
VOPCOp.V_CMPX_GE_I64: _VOPCOp_V_CMPX_GE_I64,
VOPCOp.V_CMPX_T_I64: _VOPCOp_V_CMPX_T_I64,
VOPCOp.V_CMPX_F_U64: _VOPCOp_V_CMPX_F_U64,
VOPCOp.V_CMPX_LT_U64: _VOPCOp_V_CMPX_LT_U64,
VOPCOp.V_CMPX_EQ_U64: _VOPCOp_V_CMPX_EQ_U64,
VOPCOp.V_CMPX_LE_U64: _VOPCOp_V_CMPX_LE_U64,
VOPCOp.V_CMPX_GT_U64: _VOPCOp_V_CMPX_GT_U64,
VOPCOp.V_CMPX_NE_U64: _VOPCOp_V_CMPX_NE_U64,
VOPCOp.V_CMPX_GE_U64: _VOPCOp_V_CMPX_GE_U64,
VOPCOp.V_CMPX_T_U64: _VOPCOp_V_CMPX_T_U64,
VOPCOp.V_CMPX_CLASS_F16: _VOPCOp_V_CMPX_CLASS_F16,
VOPCOp.V_CMPX_CLASS_F32: _VOPCOp_V_CMPX_CLASS_F32,
VOPCOp.V_CMPX_CLASS_F64: _VOPCOp_V_CMPX_CLASS_F64,
}
# V_WRITELANE_B32: Write scalar to specific lane's VGPR (not in PDF pseudocode)
def _VOP3Op_V_WRITELANE_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
wr_lane = S1._val & 0x1f # lane select (5 bits for wave32)
return {'vgpr_write': (wr_lane, VDST._val, S0._val & 0xffffffff)}
VOP3Op_FUNCTIONS[VOP3Op.V_WRITELANE_B32] = _VOP3Op_V_WRITELANE_B32
# V_PERM_B32: Byte permutation (not in PDF pseudocode)
# Combined 64-bit data = {S0, S1} where S1 is bytes 0-3, S0 is bytes 4-7
# S2 is selector: each byte selects which byte of combined data goes to output
def _VOP3Op_V_PERM_B32(S0, S1, S2, D0, SCC, VCC, laneId, EXEC, SIMM16, VGPR, SRC0, VDST):
from extra.assembly.amd.pcode import BYTE_PERMUTE
combined = (S0._val << 32) | S1._val # {S0, S1}
sel = S2._val
result = 0
for i in range(4):
byte_sel = (sel >> (i * 8)) & 0xff
result |= BYTE_PERMUTE(combined, byte_sel) << (i * 8)
D0.b32 = result
VOP3Op_FUNCTIONS[VOP3Op.V_PERM_B32] = _VOP3Op_V_PERM_B32
COMPILED_FUNCTIONS = {
SOP1Op: SOP1Op_FUNCTIONS,
SOP2Op: SOP2Op_FUNCTIONS,
SOPCOp: SOPCOp_FUNCTIONS,
SOPKOp: SOPKOp_FUNCTIONS,
SOPPOp: SOPPOp_FUNCTIONS,
VOP1Op: VOP1Op_FUNCTIONS,
VOP2Op: VOP2Op_FUNCTIONS,
VOP3Op: VOP3Op_FUNCTIONS,
VOP3SDOp: VOP3SDOp_FUNCTIONS,
VOP3POp: VOP3POp_FUNCTIONS,
VOPCOp: VOPCOp_FUNCTIONS,
}
def get_compiled_functions(): return COMPILED_FUNCTIONS
Reference in New Issue View Git Blame Copy Permalink