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

cuda : clear error after buffer allocation failure

Browse source
This commit is contained in:
slaren 2024-05-19 00:39:15 +02:00
parent 059031b8c4
commit 69d392ae9c
1 changed files with 2 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

2
ggml-cuda.cu
View file

@ -499,6 +499,8 @@ GGML_CALL static ggml_backend_buffer_t ggml_backend_cuda_buffer_type_alloc_buffe
void * dev_ptr; void * dev_ptr;
cudaError_t err = cudaMalloc(&dev_ptr, size); cudaError_t err = cudaMalloc(&dev_ptr, size);
if (err != cudaSuccess) { if (err != cudaSuccess) {
// clear the error
cudaGetLastError();
fprintf(stderr, "%s: allocating %.2f MiB on device %d: cudaMalloc failed: %s\n", __func__, size/1024.0/1024.0, buft_ctx->device, cudaGetErrorString(err)); fprintf(stderr, "%s: allocating %.2f MiB on device %d: cudaMalloc failed: %s\n", __func__, size/1024.0/1024.0, buft_ctx->device, cudaGetErrorString(err));
return nullptr; return nullptr;
} }