ver4a/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

cuda : enable CUDA Graph on CUDA Toolkit < 12.x (#12394)

Browse source 
* Enable CUDA Graph on CTK < 12.x

`cudaGraphExecUpdate` API was changed on 12.x. For this reason CUDA graph support was disabled on older CUDA toolkit. This change enables CUDA support in CTK version < 12.x by using older API if CTK < 12.x.

* Fix compilation errors with MUSA

* Disable CUDA Graph for MUSA
This commit is contained in:
Gaurav Garg 2025-03-17 23:55:13 +05:30 • committed by GitHub
parent 01e8f2138b
commit b1b132efcb
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
5 changed files with 11 additions and 12 deletions
Download patch file Download diff file Expand all files Collapse all files

12
ggml/src/ggml-cuda/ggml-cuda.cu
View file

@ -2610,13 +2610,15 @@ static bool is_cuda_graph_update_required(ggml_backend_cuda_context * cuda_ctx,
static void update_cuda_graph_executable(ggml_backend_cuda_context * cuda_ctx) {
#if CUDART_VERSION >= 12000
cudaGraphExecUpdateResultInfo result_info;
#ifdef __HIP_PLATFORM_AMD__
hipGraphNode_t errorNode;
hipError_t stat = hipGraphExecUpdate(cuda_ctx->cuda_graph->instance, cuda_ctx->cuda_graph->graph, &errorNode, &result_info);
#else
cudaError_t stat = cudaGraphExecUpdate(cuda_ctx->cuda_graph->instance, cuda_ctx->cuda_graph->graph, &result_info);
#endif
#else
cudaGraphNode_t errorNode;
cudaGraphExecUpdateResult result_info;
cudaError_t stat = cudaGraphExecUpdate(cuda_ctx->cuda_graph->instance, cuda_ctx->cuda_graph->graph, &errorNode, &result_info);
#endif // CUDART_VERSION >= 12000
if (stat == cudaErrorGraphExecUpdateFailure) {
#ifndef NDEBUG
GGML_LOG_DEBUG("%s: CUDA graph update failed\n", __func__);