Skip to content

cuda : fix GGML_CUDA_GRAPHS=OFF #15300

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
merged 3 commits into from
Aug 14, 2025
Merged

cuda : fix GGML_CUDA_GRAPHS=OFF #15300

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
67630cf
fix USE_CUDA_GRAPH=OFF
CISC Aug 13, 2025
f056852
check capture status
CISC Aug 13, 2025
076ebca
completely disable capturing check instead
CISC Aug 13, 2025
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
8 changes: 7 additions & 1 deletion ggml/src/ggml-cuda/mean.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -25,9 +25,12 @@ void ggml_cuda_op_mean(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {

// Special case for reducing vectors
#ifdef GGML_CUDA_USE_CUB
#ifdef USE_CUDA_GRAPH
cudaStreamCaptureStatus iscapturing;
CUDA_CHECK(cudaStreamIsCapturing(stream, &iscapturing));
#endif // USE_CUDA_GRAPH
if ((nrows == 1) &&
#ifdef USE_CUDA_GRAPH
// CUDA_GRAPHS_DISABLED
((ncols > 65536) &&
((ctx.cuda_graph->instance == nullptr) && (iscapturing == cudaStreamCaptureStatusNone) ||
Expand All @@ -38,6 +41,9 @@ void ggml_cuda_op_mean(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
!((ctx.cuda_graph->instance == nullptr) && (iscapturing == cudaStreamCaptureStatusNone) ||
ctx.cuda_graph->disable_due_to_gpu_arch || ctx.cuda_graph->disable_due_to_too_many_updates ||
ctx.cuda_graph->disable_due_to_failed_graph_capture))) {
#else
(ncols > 65536)) {
#endif // USE_CUDA_GRAPH
// Single row - use device-wide reduction
size_t tmp_size = 0;
ggml_cuda_pool & pool = ctx.pool();
Expand All @@ -51,7 +57,7 @@ void ggml_cuda_op_mean(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
divide_by_count<float><<<1, 1, 0, stream>>>(dst_d, ncols);
return;
}
#endif
#endif // GGML_CUDA_USE_CUB

const dim3 block_nums(nrows, 1, 1);

Expand Down
Loading