Skip to content

Commit 0f2dc34

Browse files
Fix unintended change
1 parent 0cc5c95 commit 0f2dc34

File tree

1 file changed

+2
-2
lines changed

1 file changed

+2
-2
lines changed

csrc/ops.cu

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -824,8 +824,8 @@ template <typename T> void gemm_4bit_inference(int m, int n, int k, T * A, unsi
824824
template <typename T, int BITS> void gemm_4bit_inference_naive(int m, int n, int k, T * A, unsigned char* B, float *absmax, float *datatype, T * out, int lda, int ldb, int ldc, int blocksize, cudaStream_t stream)
825825
{
826826

827-
int num_blocks = (m+7)/8;
828-
kgemm_4bit_inference_naive<T, 256, BITS><<< num_blocks, 256, 0, stream>>>(m, n, k, A, B, absmax, datatype, out, lda, ldb, ldc, blocksize);
827+
int num_blocks = (m+3)/4;
828+
kgemm_4bit_inference_naive<T, 128, BITS><<< num_blocks, 128, 0, stream>>>(m, n, k, A, B, absmax, datatype, out, lda, ldb, ldc, blocksize);
829829
CUDA_CHECK_RETURN(cudaPeekAtLastError());
830830
}
831831

0 commit comments

Comments
 (0)