Skip to content

Commit d883ac3

Browse files
authored
Match cuBLAS sync behavior (#169)
1 parent bc4ea53 commit d883ac3

1 file changed

Lines changed: 4 additions & 0 deletions

File tree

src/hydrogen/device/rocBLAS_API.cpp

Lines changed: 4 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -75,6 +75,10 @@ class ResultMgr
7575
H_CHECK_HIP(hipFreeAsync(device_, stream));
7676
#endif // HYDROGEN_HAVE_CUB
7777

78+
// Sync stream to match cuBLAS behavior (cuBLAS docs here:
79+
// https://docs.nvidia.com/cuda/cublas/#scalar-parameters)
80+
H_CHECK_HIP(hipStreamSynchronize(stream));
81+
7882
// Reset pointer mode
7983
H_CHECK_ROCBLAS(rocblas_set_pointer_mode(handle_, rocblas_pointer_mode_host));
8084
}

0 commit comments

Comments
 (0)