Skip to content

HIP/CUDA: Fix reduction in mmf getting tile widths not devisiable by 64 in mul_mat_id on wave64 devices #30836

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
ggerganov merged 1 commit into from
Oct 20, 2026
Merged

HIP/CUDA: Fix reduction in mmf getting tile widths not devisiable by 64 in mul_mat_id on wave64 devices #30836

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 5 additions & 2 deletions ggml/src/ggml-cuda/ggml-cuda.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2014,6 +2014,8 @@ static bool ggml_cuda_mul_mat_id_needs_sync(const ggml_tensor * dst, const int c
const ggml_tensor * src0 = dst->src[0];
const ggml_tensor * src1 = dst->src[1];

const int warp_size = ggml_cuda_info().devices[ggml_cuda_get_device()].warp_size;

if (src1->type != GGML_TYPE_F32 || dst->type != GGML_TYPE_F32) {
return true;
}
Expand All @@ -2032,7 +2034,7 @@ static bool ggml_cuda_mul_mat_id_needs_sync(const ggml_tensor * dst, const int c
return false;
}

if (ggml_cuda_should_use_mmf(src0->type, cc, WARP_SIZE, src0->ne, src0->nb, src1->ne[2], /*mul_mat_id=*/true)) {
if (ggml_cuda_should_use_mmf(src0->type, cc, warp_size, src0->ne, src0->nb, src1->ne[2], /*mul_mat_id=*/true)) {
return false;
}

Expand All @@ -2050,6 +2052,7 @@ static void ggml_cuda_mul_mat_id(ggml_backend_cuda_context & ctx, ggml_tensor *
GGML_TENSOR_BINARY_OP_LOCALS

const int cc = ggml_cuda_info().devices[ggml_cuda_get_device()].cc;
const int warp_size = ggml_cuda_info().devices[ggml_cuda_get_device()].warp_size;

// [TAG_MUL_MAT_ID_CUDA_GRAPHS]
if (src1->type == GGML_TYPE_F32 && dst->type == GGML_TYPE_F32) {
Expand All @@ -2074,7 +2077,7 @@ static void ggml_cuda_mul_mat_id(ggml_backend_cuda_context & ctx, ggml_tensor *
return;
}

if (ggml_cuda_should_use_mmf(src0->type, cc, WARP_SIZE, src0->ne, src0->nb, src1->ne[2], /*mul_mat_id=*/true)) {
if (ggml_cuda_should_use_mmf(src0->type, cc, warp_size, src0->ne, src0->nb, src1->ne[2], /*mul_mat_id=*/true)) {
ggml_cuda_mul_mat_f(ctx, src0, src1, ids, dst);
return;
}
Expand Down