github/ROCm
1
1
Fork 0
mirror of https://github.com/ROCm/ROCm.git synced 2026-04-05 03:01:17 -04:00
Code Issues Packages Projects Releases Wiki Activity

Lit test to check for illegal st.shared.b1 llvmir (#2387)

Browse Source
This commit is contained in:
Tori Baker
2023-09-26 19:12:32 +02:00
committed by GitHub
parent 2d28b09319
commit bf3171f5c7
1 changed files with 20 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

20
test/Conversion/tritongpu_to_llvm.mlir
View File

@@ -1435,3 +1435,23 @@ module attributes {"triton_gpu.compute-capability" = 80 : i32, "triton_gpu.num-c
tt.return
}
}
// -----
// CHECK-LABEL: copyitem
// CHECK: st.shared.b8
// CHECK: ld.shared.b8
// CHECK-NOT: st.shared.b1
// CHECK-NOT: ld.shared.b1
#blocked = #triton_gpu.blocked<{sizePerThread = [1, 1], threadsPerWarp = [4, 8], warpsPerCTA = [1, 4], order = [0, 1], CTAsPerCGA = [1, 1], CTASplitNum = [1, 1], CTAOrder = [0, 1]}>
module attributes {"triton_gpu.compute-capability" = 80 : i32, "triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 4 : i32, "triton_gpu.threads-per-warp" = 32 : i32} {
tt.func public @copyitem() attributes {noinline = false} {
%cst = arith.constant dense<true> : tensor<4x1xi1, #blocked>
%0 = "tt.reduce"(%cst) <{axis = 1 : i32}> ({
^bb0(%arg0: i1, %arg1: i1):
%1 = arith.ori %arg0, %arg1 : i1
tt.reduce.return %1 : i1
}) : (tensor<4x1xi1, #blocked>) -> tensor<4xi1, #triton_gpu.slice<{dim = 1, parent = #blocked}>>
tt.return
}
}