happyz
/
llama.cpp
mirror of https://github.com/ggerganov/llama.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

BUILD: update build doc, add cmake preset, add CACHE_DIR env var

This commit is contained in:
Yu, Zijun 2025-04-29 14:31:35 +08:00 committed by Mustafa Cavus
parent d3bdca25bd
commit 0a8cc9ab03
2 changed files with 25 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

20
CMakePresets.json
View File

@ -1,6 +1,26 @@
{ {
"version": 4, "version": 4,
"configurePresets": [ "configurePresets": [
{
"name": "ReleaseOV",
"generator": "Ninja",
"binaryDir": "${sourceDir}/build/${presetName}",
"installDir": "${sourceDir}/build/install/${presetName}",
"cacheVariables": {
"CMAKE_BUILD_TYPE": "Release",
"GGML_OPENVINO": true,
"OpenVINO_DIR": "$env{OPENVINO_LLAMA_PATH}/build/Release"
}
},
{
"name": "ReleaseCPU",
"generator": "Ninja",
"binaryDir": "${sourceDir}/build/${presetName}",
"installDir": "${sourceDir}/build/install/${presetName}",
"cacheVariables": {
"CMAKE_BUILD_TYPE": "Release"
}
},
{ {
"name": "base", "name": "base",
"hidden": true, "hidden": true,

8
ggml/src/ggml-openvino/utils.cpp
View File

@ -37,7 +37,6 @@ std::map<std::string, void*> get_ggml_graph_output_dst(std::shared_ptr<GgmlOvDec
} }
static ov::frontend::FrontEnd::Ptr get_ggml_frontend() { static ov::frontend::FrontEnd::Ptr get_ggml_frontend() {
ov::frontend::FrontEnd::Ptr front_end = nullptr;
auto fem = ov::frontend::FrontEndManager(); auto fem = ov::frontend::FrontEndManager();
std::string fe_so_path; std::string fe_so_path;
#ifdef GGML_OV_FRONTEND #ifdef GGML_OV_FRONTEND
@ -52,6 +51,10 @@ enum ggml_status openvino_frontend_compute(ggml_backend_t backend, struct ggml_c
auto start_time = ggml_time_us(); auto start_time = ggml_time_us();
static ov::Core core; static ov::Core core;
auto* cache_dir = getenv("GGML_OPENVINO_CACHE_DIR");
if (cache_dir) {
core.set_property(ov::cache_dir(cache_dir));
}
// auto devices = core.get_available_devices(); // auto devices = core.get_available_devices();
static auto front_end = get_ggml_frontend(); static auto front_end = get_ggml_frontend();
@ -82,8 +85,7 @@ enum ggml_status openvino_frontend_compute(ggml_backend_t backend, struct ggml_c
GGML_LOG_ERROR("Model is not converted \n"); GGML_LOG_ERROR("Model is not converted \n");
} }
ov::CompiledModel compiled_model = ov::CompiledModel compiled_model = core.compile_model(model, "CPU");
core.compile_model(model, "CPU", ov::device::properties("CPU", ov::cache_dir("/tmp/ov_cache")));
auto compile_end_time = ggml_time_us(); auto compile_end_time = ggml_time_us();
ov::InferRequest infer_request = compiled_model.create_infer_request(); ov::InferRequest infer_request = compiled_model.create_infer_request();