github/tinygrad
1
1
Fork 0
mirror of https://github.com/tinygrad/tinygrad.git synced 2026-04-07 03:00:26 -04:00
Code Issues Packages Projects Releases Wiki Activity

nv limit mappings debug (#7215)

Browse Source
This commit is contained in:
nimlgen
2024-10-22 16:41:43 +03:00
committed by GitHub
parent 4013c9848c
commit cef7078c14
1 changed files with 3 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
tinygrad/runtime/ops_nv.py
View File

@@ -571,9 +571,10 @@ class NVDevice(HCQCompiled):
for i in range(mmu_info.count):
pfinfo = mmu_info.mmuFaultInfoList[i]
report += [f"MMU fault: 0x{pfinfo.faultAddress:X} | {NV_PFAULT_FAULT_TYPE[pfinfo.faultType]} | {NV_PFAULT_ACCESS_TYPE[pfinfo.accessType]}"]
report += ["GPU mappings:\n"+"\n".join([f"\t0x{x:X} - 0x{x+y-1:X} | {self._debug_mappings[(x,y)]}" for x,y in sorted(self._debug_mappings)])]
if DEBUG >= 5:
report += ["GPU mappings:\n"+"\n".join(f"\t0x{x:X} - 0x{x+y-1:X} | {self._debug_mappings[(x,y)]}" for x,y in sorted(self._debug_mappings))]
else:
for i, e in enumerate(sm_errors.smErrorStateArray):
if e.hwwGlobalEsr or e.hwwWarpEsr: report += [f"SM{i} fault: esr={e.hwwGlobalEsr} warp_esr={e.hwwWarpEsr} warp_pc={e.hwwWarpEsrPc64}"]
if e.hwwGlobalEsr or e.hwwWarpEsr: report += [f"SM {i} fault: esr={e.hwwGlobalEsr} warp_esr={e.hwwWarpEsr} warp_pc={e.hwwWarpEsrPc64}"]
raise RuntimeError("\n".join(report))