CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-03 09:22:01 +00:00
Code Issues Packages Projects Releases Wiki Activity

initial implementation of delayed graph allocation

Browse Source
This commit is contained in:
slaren
2023-07-20 15:57:48 +02:00
parent cb205c0d13
commit de69f8f20d
6 changed files with 165 additions and 87 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
ggml-cuda.cu
View File

@@ -1752,6 +1752,8 @@ static void ggml_backend_cuda_get_tensor_async(ggml_backend * backend, const ggm
//ggml_backend_cuda_context * cuda_ctx = (ggml_backend_cuda_context *)backend->context;
//printf("get tensor %s %p\n", tensor->name, tensor->data);
CUDA_CHECK(cudaMemcpyAsync(data, (const char*)tensor->data + offset, size, cudaMemcpyDeviceToHost, g_cudaStream_main));
UNUSED(backend);