Lumpiasty/llama.cpp
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

ggml-cuda: refactor cuda graph usage (#18637)

Browse Source 
* ggml-cuda: refactor cuda graph usage

* use is_enabled() instead of enabled
This commit is contained in:
Aman Gupta
2026-01-06 23:48:45 +08:00
committed by GitHub
parent 968929528c
commit 090b137e56
3 changed files with 72 additions and 96 deletions
Download Patch File Download Diff File Expand all files Collapse all files
ggml/src/ggml-cuda/mean.cu
+2 -4
View File
@@ -34,13 +34,11 @@ void ggml_cuda_op_mean(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
// CUDA_GRAPHS_DISABLED
((ncols > 65536) &&
((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)) ||
ctx.cuda_graph->is_enabled())) ||
// CUDA_GRAPHS ENABLED
((ncols > 32768) &&
!((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))) {
ctx.cuda_graph->is_enabled()))) {
#else
(ncols > 65536)) {
#endif // USE_CUDA_GRAPH