nicolas
/
ik_llama_opt
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Do not create CUDA graphs when disabled (#1702)

This commit is contained in:
Kawrakow 2026-04-27 16:36:12 +02:00 committed by GitHub
parent 0a167082a3
commit d6f3e4e28f
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
ggml/src/ggml-cuda.cu
View File

@ -4283,7 +4283,7 @@ static void evaluate_and_capture_cuda_graph(ggml_backend_cuda_context * cuda_ctx
[[maybe_unused]] const bool integrated = false; //ggml_cuda_info().devices[cuda_ctx->device].integrated; [[maybe_unused]] const bool integrated = false; //ggml_cuda_info().devices[cuda_ctx->device].integrated;
#ifdef USE_CUDA_GRAPH #ifdef USE_CUDA_GRAPH
auto graph = ggml_cuda_get_graph(*cuda_ctx, ggml_cuda_graph_get_key(cgraph)); auto graph = use_cuda_graph ? ggml_cuda_get_graph(*cuda_ctx, ggml_cuda_graph_get_key(cgraph)) : nullptr;
#endif #endif
//printf("======================== %s: graph with %d nodes on device %d. time = %ld\n", __func__, cgraph->n_nodes, cuda_ctx->device, ggml_time_us()); //printf("======================== %s: graph with %d nodes on device %d. time = %ld\n", __func__, cgraph->n_nodes, cuda_ctx->device, ggml_time_us());