Skip to content

Commit 1af6853

Browse files
JohannesGaesslerarthw
authored andcommitted
CUDA: fix MMV kernel being used for FP16 src1 (ggml-org#10357)
1 parent 7af08cb commit 1af6853

File tree

1 file changed

+4
-2
lines changed

1 file changed

+4
-2
lines changed

ggml/src/ggml-cuda/ggml-cuda.cu

Lines changed: 4 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -1760,11 +1760,13 @@ static void ggml_cuda_mul_mat(ggml_backend_cuda_context & ctx, const ggml_tensor
17601760
//printf("src0 is contiguous %d, transposed %d, type = %s, name = %s\n", ggml_is_contiguous(src0), ggml_is_transposed(src0), ggml_type_name(src0->type), src0->name);
17611761
//printf("src1 is contiguous %d, transposed %d, type = %s, name = %s\n", ggml_is_contiguous(src1), ggml_is_transposed(src1), ggml_type_name(src1->type), src1->name);
17621762

1763-
if (!split && src0->type == GGML_TYPE_F16 && src1->ne[1] == 1 && dst->ne[3] == 1 && (src0->ne[1] < MMV_MAX_ROWS || any_gpus_without_fp16_mma)) {
1763+
if (!split && use_mul_mat_vec && dst->ne[3] == 1 && (src0->ne[1] < MMV_MAX_ROWS || any_gpus_without_fp16_mma)) {
1764+
// the custom F16 vector kernel can be used over batched cuBLAS GEMM
1765+
// but this is only faster for GPUs without tensor cores or with a thin src0 matrix (particularly KQV in attention)
17641766
ggml_cuda_mul_mat_vec(ctx, src0, src1, dst);
17651767
} else if (!split && src0->type == GGML_TYPE_F16 && (src1->type == GGML_TYPE_F16 || !any_gpus_with_slow_fp16)
17661768
&& !ggml_is_transposed(src0) && !ggml_is_transposed(src1) && src1->ne[2]*src1->ne[3] > 1) {
1767-
// KQ + KQV multi-batch without FlashAttention
1769+
// general KQ + KQV multi-batch without FlashAttention
17681770
ggml_cuda_mul_mat_batched_cublas(ctx, src0, src1, dst);
17691771
} else if (use_mul_mat_vec) {
17701772
ggml_cuda_op_mul_mat(ctx, src0, src1, dst, ggml_cuda_op_mul_mat_vec, nullptr);

0 commit comments

Comments
 (0)