microsoft/mscclpp
1
0
Fork 0
mirror of https://github.com/microsoft/mscclpp.git synced 2026-05-22 22:08:28 +00:00
Code Issues Packages Projects Releases Wiki Activity

Do not check value of __HIP_PLATFORM_AMD__ (#240)

Browse Source 
According to the
[document](https://rocm.docs.amd.com/projects/HIP/en/docs-6.0.0/user_guide/hip_porting_guide.html#compiler-defines-summary),
`__HIP_PLATFORM_AMD__` is effective only by definition.
This commit is contained in:
Changho Hwang
2023-12-25 13:51:18 +08:00
committed by GitHub
parent 5fa5bd2706
commit 70e28b3c76
6 changed files with 11 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
test/allgather_test_cpp.cu
View File

@@ -74,7 +74,7 @@ __device__ void localAllGather(DeviceHandle<mscclpp::SimpleProxyChannel> proxyCh
if ((remoteRank % nranksPerNode) == ((rank - i + nranksPerNode) % nranksPerNode)) {
if ((threadIdx.x % 32) == 0) proxyChan.wait();
}
#if defined(__HIP_PLATFORM_AMD__) && (__HIP_PLATFORM_AMD__ == 1)
#if defined(__HIP_PLATFORM_AMD__)
// NOTE: we actually need a group barrier here for better performance, but __syncthreads() is still correct.
__syncthreads();
#else