Mirrors/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-10-30 06:30:15 +01:00
Code Issues Packages Projects Releases Wiki Activity

OpenCL: Add release memory (#1741)

Browse Source 
* Add opencl release memory

* Rename function name
This commit is contained in:
Robert Sung-wook Shin 2023-06-10 01:24:40 +09:00 committed by GitHub
parent ae9663f188
commit 98ed165574
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
3 changed files with 16 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
ggml-opencl.cpp
View File

@ -662,6 +662,15 @@ static void ggml_cl_pool_free(cl_mem mem, size_t size) {
clReleaseMemObject(mem);
}
void ggml_cl_free_data(const struct ggml_tensor* tensor) {
if (tensor->backend != GGML_BACKEND_GPU) {
return;
}
cl_mem mem = (cl_mem)tensor->data;
clReleaseMemObject(mem);
}
static cl_int ggml_cl_h2d_tensor_2d(cl_command_queue queue, cl_mem dst, size_t offset, const struct ggml_tensor * src, uint64_t i3, uint64_t i2, cl_event* ev) {
cl_int err;
const uint64_t ne0 = src->ne[0];

2
ggml-opencl.h
View File

@ -16,6 +16,8 @@ void ggml_cl_mul_mat(const struct ggml_tensor * src0, const struct ggml_tensor
void * ggml_cl_host_malloc(size_t size);
void ggml_cl_host_free(void * ptr);
void ggml_cl_free_data(const struct ggml_tensor* tensor);
void ggml_cl_transform_tensor(struct ggml_tensor * tensor);
void ggml_cl_load_data(const char * fname, struct ggml_tensor * tensor, size_t offset);

6
llama.cpp
View File

@ -210,7 +210,11 @@ struct llama_model {
for (size_t i = 0; i < tensors_by_name.size(); ++i) {
ggml_cuda_free_data(tensors_by_name[i].second);
}
#endif // GGML_USE_CUBLAS
#elif defined(GGML_USE_CLBLAST)
for (size_t i = 0; i < tensors_by_name.size(); ++i) {
ggml_cl_free_data(tensors_by_name[i].second);
}
#endif
}
};