Skip to content

Commit 4adb1d6

Browse files
committed
cuda : add comment
1 parent dd86df8 commit 4adb1d6

File tree

1 file changed

+1
-0
lines changed

1 file changed

+1
-0
lines changed

ggml-cuda.cu

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -7849,6 +7849,7 @@ static void ggml_cuda_mul_mat(const ggml_tensor * src0, const ggml_tensor * src1
78497849
#endif // GGML_CUDA_FORCE_DMMV
78507850

78517851
if (use_mul_mat_vec_q) {
7852+
// NOTE: this kernel does not support ggml_nrows(src1) > 1
78527853
ggml_cuda_op_mul_mat(src0, src1, dst, ggml_cuda_op_mul_mat_vec_q, true);
78537854
} else {
78547855
ggml_cuda_op_mul_mat(src0, src1, dst, ggml_cuda_op_dequantize_mul_mat_vec, false);

0 commit comments

Comments
 (0)