Skip to content

CUDA: only disable CUDA graphs when mul_mat_id actually needs a stream sync #26802

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
Aug 11, 2026
+38 −5
Merged

CUDA: only disable CUDA graphs when mul_mat_id actually needs a stream sync #26802

Show file tree
Hide file tree
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
39 changes: 34 additions & 5 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 @@ -1865,6 +1865,37 @@ static void ggml_cuda_mul_mat(ggml_backend_cuda_context & ctx, const ggml_tensor
ggml_cuda_mul_mat_cublas(ctx, src0, src1, dst);
}

// returns true when ggml_cuda_mul_mat_id takes the fallback path that requires stream synchronization
// [TAG_MUL_MAT_ID_CUDA_GRAPHS]
static bool ggml_cuda_mul_mat_id_needs_sync(const ggml_tensor * dst, const int cc) {
const ggml_tensor * src0 = dst->src[0];
const ggml_tensor * src1 = dst->src[1];

if (src1->type != GGML_TYPE_F32 || dst->type != GGML_TYPE_F32) {
return true;
}

if (dst->ne[2] <= MMVQ_MAX_BATCH_SIZE) {
if (ggml_is_quantized(src0->type)) {
if (dst->ne[2] <= get_mmvq_mmid_max_batch(src0->type, cc)) {
return false;
}
} else if (GGML_CUDA_CC_IS_AMD(cc)) {
return false;
}
}

if (ggml_cuda_should_use_mmq(src0->type, cc, src1->ne[2], /*n_experts=*/src0->ne[2])) {
return false;
}

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

return true;
}

static void ggml_cuda_mul_mat_id(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
const ggml_tensor * src0 = dst->src[0];
const ggml_tensor * src1 = dst->src[1];
Expand Down Expand Up @@ -1907,7 +1938,7 @@ static void ggml_cuda_mul_mat_id(ggml_backend_cuda_context & ctx, ggml_tensor *
}

// note: this path should not be reached when recording CUDA graphs, because it requires stream synchronization
// TODO: add asserts to verify this. should work with CUDA, HIP, etc.
GGML_ASSERT(ggml_cuda_mul_mat_id_needs_sync(dst, cc));
cudaStream_t stream = ctx.stream();

GGML_ASSERT(nb12 % nb11 == 0);
Expand Down Expand Up @@ -2522,10 +2553,8 @@ static bool ggml_cuda_graph_check_compability(ggml_cgraph * cgraph) {
// [TAG_MUL_MAT_ID_CUDA_GRAPHS]
if (node->op == GGML_OP_MUL_MAT_ID) {
const int cc = ggml_cuda_info().devices[ggml_cuda_get_device()].cc;
const int mmvq_mmid_max = get_mmvq_mmid_max_batch(node->src[0]->type, cc);
if (!ggml_is_quantized(node->src[0]->type) || node->ne[2] > mmvq_mmid_max) {
// under these conditions, the mul_mat_id operation will need to synchronize the stream, so we cannot use CUDA graphs
// TODO: figure out a way to enable for larger batch sizes, without hurting performance
if (ggml_cuda_mul_mat_id_needs_sync(node, cc)) {
// the mul_mat_id fallback path synchronizes the stream, so we cannot use CUDA graphs
// ref: https://github.com/ggml-org/llama.cpp/pull/18958
use_cuda_graph = false;
#ifndef NDEBUG
Expand Down
4 changes: 4 additions & 0 deletions tests/test-backend-ops.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9004,6 +9004,8 @@ static std::vector<std::unique_ptr<test_case>> make_test_cases_eval() {
test_cases.emplace_back(new test_mul_mat(GGML_TYPE_F32, GGML_TYPE_F32, 64, 128, k, {12,1}, {1,1}));
test_cases.emplace_back(new test_mul_mat_id(GGML_TYPE_F16, GGML_TYPE_F32, 16, 16, false, 50, 200, k));
test_cases.emplace_back(new test_mul_mat_id(GGML_TYPE_F16, GGML_TYPE_F32, 16, 16, true, 50, 200, k));
test_cases.emplace_back(new test_mul_mat_id(GGML_TYPE_BF16, GGML_TYPE_F32, 16, 16, false, 50, 200, k));
test_cases.emplace_back(new test_mul_mat_id(GGML_TYPE_BF16, GGML_TYPE_F32, 16, 16, true, 50, 200, k));
test_cases.emplace_back(new test_mul_mat_id(GGML_TYPE_F32, GGML_TYPE_F32, 16, 16, false, 50, 200, k));
test_cases.emplace_back(new test_mul_mat_id(GGML_TYPE_F32, GGML_TYPE_F32, 16, 16, true, 50, 200, k));
}
Expand Down Expand Up @@ -9035,6 +9037,8 @@ static std::vector<std::unique_ptr<test_case>> make_test_cases_eval() {
test_cases.emplace_back(new test_mul_mat_id(GGML_TYPE_F16, GGML_TYPE_F32, 16, 16, b, 32, 1024, 16));
test_cases.emplace_back(new test_mul_mat_id(GGML_TYPE_F16, GGML_TYPE_F32, 2, 2, b, 32, 8192, 64));
test_cases.emplace_back(new test_mul_mat_id(GGML_TYPE_F16, GGML_TYPE_F32, 16, 16, b, 50, 200, 64));
test_cases.emplace_back(new test_mul_mat_id(GGML_TYPE_BF16, GGML_TYPE_F32, 16, 16, b, 32, 1024, 16));
test_cases.emplace_back(new test_mul_mat_id(GGML_TYPE_BF16, GGML_TYPE_F32, 16, 16, b, 50, 200, 64));
}

test_cases.emplace_back(new test_mul_mat_id(GGML_TYPE_F16, GGML_TYPE_F32, 1, 1, false, 8, 16, 1));
Expand Down
Loading