vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Removed redundant cublasSetStream

Browse source
This commit is contained in:
Oleksii Maryshchenko 2023-11-02 12:32:01 +01:00
parent 7e6f41327a
commit 587ff3bf1a
1 changed files with 0 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

1
ggml-cuda.cu
View file

@ -7387,7 +7387,6 @@ static void ggml_cuda_mul_mat_mat_batched_cublas(const ggml_tensor * src0, const
dst->nb[2], dst->nb[3],
r2, r3);
CUDA_CHECK(cudaGetLastError());
CUBLAS_CHECK(cublasSetStream(g_cublas_handles[id], main_stream));
CUBLAS_CHECK(
cublasGemmBatchedEx(g_cublas_handles[id], CUBLAS_OP_T, CUBLAS_OP_N,
ne01, ne11, ne10,