Skip to content

Commit 8ffe4de

Browse files
authored
[OpenMP] [amdgpu] Enable small blocksize for generic SPMD kernels. (llvm#4304)
2 parents 5bf4832 + 2fb4778 commit 8ffe4de

File tree

1 file changed

+1
-1
lines changed
  • offload/plugins-nextgen/amdgpu/src

1 file changed

+1
-1
lines changed

offload/plugins-nextgen/amdgpu/src/rtl.cpp

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -3160,7 +3160,7 @@ struct AMDGPUDeviceTy : public GenericDeviceTy, AMDGenericDeviceTy {
31603160
OMPX_AdjustNumTeamsForXteamRedSmallBlockSize(
31613161
"LIBOMPTARGET_AMDGPU_ADJUST_XTEAM_RED_TEAMS", 1),
31623162
OMPX_GenericSpmdUseSmallBlockSize(
3163-
"LIBOMPTARGET_AMDGPU_GENERIC_SPMD_USE_SMALL_BLOCKSIZE", 0),
3163+
"LIBOMPTARGET_AMDGPU_GENERIC_SPMD_USE_SMALL_BLOCKSIZE", 1),
31643164
OMPX_MaxAsyncCopyBytes("LIBOMPTARGET_AMDGPU_MAX_ASYNC_COPY_BYTES",
31653165
64 * 1024),
31663166
OMPX_InitialNumSignals("LIBOMPTARGET_AMDGPU_NUM_INITIAL_HSA_SIGNALS",

0 commit comments

Comments
 (0)