llama-cpp-turboquant/ggml/src/ggml-backend-reg.cpp
Zijun Yu 9789c4ecdc
ggml : add OpenVINO backend (#15307)
* Update build doc

* Add cgraph tensor output name to OV op name

* Update openvino build instructions

* Add initial NPU support

* draft NPU support version 2: prefill + kvcache

* NPU support version 2: prefill + kvcache

* Change due to ggml cgraph changes, not correct yet

* Change due to ggml cgraph changes, llama-3.2 CPU work

* Add AMD64 to CMakeLists

* Change due to ggml cgraph changes, all device work

* Refactor: clean, fix warning

* Update clang-format

* Statful transformation for CPU GPU

* Add SwiGLU

* Fuse to SDPA

* Replace Concat with Broadcast in MulMat for GQA

* Pull out indices creation for kv cache update

* Refactor: remove past_token_len from extra_inputs

* Fix Phi3 SwiGLU and SoftMax

* Pull out sin cos from rope

* Reduce memory: free ov weights node after graph conversion

* Fix CPY due to cgraph change

* Added OpenVINO CI/CD. Updated docs

* Fix llama-cli

* Fix Phi3 ROPE; Add test-backend-ops

* Fix NPU

* Fix llama-bench; Clang-format

* Fix llama-perplexity

* temp. changes for mark decomp

* matmul in fp32

* mulmat input conversion fix

* mulmat type conversion update

* add mark decomp pass

* Revert changes in fuse_to_sdpa

* Update build.md

* Fix test-backend-ops

* Skip test-thread-safety; Run ctest only in ci/run.sh

* Use CiD for NPU

* Optimize tensor conversion, improve TTFT

* Support op SET_ROWS

* Fix NPU

* Remove CPY

* Fix test-backend-ops

* Minor updates for raising PR

* Perf: RMS fused to OV internal RMS op

* Fix after rebasing

- Layout of cache k and cache v are unified: [seq, n_head, head_size]
- Add CPY and FLASH_ATTN_EXT, flash attn is not used yet
- Skip test-backend-ops due to flash attn test crash
- Add mutex around graph conversion to avoid test-thread-safety fali in the future
- Update NPU config
- Update GPU config to disable SDPA opt to make phi-3 run

* Change openvino device_type to GPU; Enable flash_attn

* Update supports_buft and supports_op for quantized models

* Add quant weight conversion functions from genai gguf reader

* Quant models run with accuracy issue

* Fix accuracy: disable cpu_repack

* Fix CI; Disable test-backend-ops

* Fix Q4_1

* Fix test-backend-ops: Treat quantized tensors as weights

* Add NPU Q4_0 support

* NPU perf: eliminate zp

* Dequantize q4_1 q4_k q6_k for NPU

* Add custom quant type: q8_1_c, q4_0_128

* Set m_is_static=false as default in decoder

* Simpilfy translation of get_rows

* Fix after rebasing

* Improve debug util; Eliminate nop ReshapeReshape

* STYLE: make get_types_to_requant a function

* Support BF16 model

* Fix NPU compile

* WA for npu 1st token acc issue

* Apply EliminateZP only for npu

* Add GeGLU

* Fix Hunyuan

* Support iSWA

* Fix NPU accuracy

* Fix ROPE accuracy when freq_scale != 1

* Minor: not add attention_size_swa for non-swa model

* Minor refactor

* Add Q5_K to support phi-3-q4_k_m

* Requantize Q6_K (gs16) to gs32 on GPU

* Fix after rebasing

* Always apply Eliminate_ZP to fix GPU compile issue on some platforms

* kvcachefusion support

* env variable GGML_OPENVINO_DISABLE_SDPA_OPTIMIZATION added

* Fix for Phi3

* Fix llama-cli (need to run with --no-warmup)

* Fix add_sliced_mask; Revert mulmat, softmax; Remove input attention_size, iSWA model not working

* fix after rebasing

* Fix llama-3-8b and phi3-mini q4_0 NPU

* Update to OV-2025.3 and CMakeLists.txt

* Add OV CI cache

* Apply CISC review and update CI to OV2025.3

* Update CI to run OV dep install before build

* Update OV dockerfile to use OV2025.3 and update build docs

* Style: use switch in supports_ops

* Style: middle ptr and ref align, omit optional struct keyword

* NPU Unify PD (#14)

* Stateless. Fix llama-cli llama-server

* Simplify broadcast op in attention

* Replace get_output_tensor+memcpy with set_output_tensor

* NPU unify PD. Unify dynamic and static dims

* Clean placeholders in ggml-openvino.cpp

* NPU unify PD (handled internally)

* change graph to 4d, support multi sequences

* Fix llama-bench

* Fix NPU

* Update ggml-decoder.cpp

Hitting error while compiling on windows:

error C3861: 'unsetenv': identifier not found

Reason: unsetenv() is a POSIX function; it doesn’t exist on Windows. Visual Studio (MSVC) won’t recognize it.

Proposed fix: Use _putenv_s() (Windows equivalent)
This is supported by MSVC and achieves the same effect: it removes the environment variable from the process environment.

This keeps cross-platform compatibility.

* Update ggml-decoder.cpp

* Update ggml-decoder.cpp

* Update ggml-decoder.cpp

* Update ggml-decoder.cpp

* Update ggml-decoder.cpp

* Remove the second decoder for node. Moving the function into the model decoder

* Fix error for naive

* NPU prefill chunking

* NPU fix llama-bench

* fallback naive run with accuracy issue

* NPU support llma-perplexity -b 512 --no-warmup

* Refactor: split ov_graph_compute for dynamic and static

* remove unused API GgmlOvDecoder::get_output_stride(const std::string & name)

* minor update due to ov 2025.4

* remove unused API GgmlOvDecoder::get_output_names()

* remove unused API get_output_shape(const std::string & name)

* Modified API GgmlOvDecoder::get_output_type(const std::string & name)

* Removed API GgmlOvDecoder::get_output_op_params(const std::string & name)

* Removed API get_output_ggml_tensor(const std::string & name)

* Removed API m_outputs

* Removed m_output_names

* Removed API GgmlOvDecoder::get_input_names()

* Removed API GgmlOvDecoder::get_input_stride(const std::string& name)

* Removed API get_input_type

* Removed API get_input_type

* Removed API GgmlOvDecoder::get_input_shape(const std::string & name)

* Removed API GgmlOvDecoder::get_input_op_params(const std::string & name)

* Fix error for decoder cache

* Reuse cached decoder

* GPU remove Q6_K requantization

* NPU fix wrong model output shape

* NPU fix q4 perf regression

* Remove unused variable nodes

* Fix decoder can_reuse for llama-bench

* Update build.md for Windows

* backend buffer: allocate on host

* Use shared_buffer for GPU NPU; Refactor

* Add ov_backend_host_buffer; Use cached remote context

* Put kvcache on GPU

* Use ggml_aligned_malloc

* only use remote tensor for kvcache

* only use remote tensor for kvcache for GPU

* FIX: use remote tensor from singleton

* Update build.md to include OpenCL

* NPU always requant to q4_0_128

* Optimize symmetric quant weight extraction: use single zp

* Use Q8_0_C in token embd, lm_head, and for 5 and 6 bits quant

* Update build.md

* Support -ctk f32

* Initial stateful graph support

* Update ggml/src/ggml-openvino/ggml-decoder.cpp

Co-authored-by: Yamini Nimmagadda <yamini.nimmagadda@intel.com>

* code cleanup

* npu perf fix

* requant to f16 for Q6 embed on NPU

* Update ggml/src/ggml-openvino/ggml-decoder.cpp

* Update ggml/src/ggml-openvino/ggml-openvino-extra.cpp

* Create OPENVINO.md in llama.cpp backend docs

* Update OPENVINO.md

* Update OPENVINO.md

* Update OPENVINO.md

* Update build.md

* Update OPENVINO.md

* Update OPENVINO.md

* Update OPENVINO.md

* kq_mask naming fix

* Syntax correction for workflows build file

* Change ov backend buffer is_host to false

* Fix llama-bench -p -n where p<=256

* Fix --direct-io 0

* Don't put kvcache on GPU in stateful mode

* Remove hardcode names

* Fix stateful shapes

* Simplification for stateful and update output shape processing

* Remove hardcode names

* Avoid re-compilation in llama-bench

* Extract zp directly instead of bias

* Refactor weight tensor processing

* create_weight_node accept non-ov backend buffer

* remove changes in llama-graph.cpp

* stateful masking fix (#38)

Fix for stateful accuracy issues and cl_out_of_resources error in stateful GPU with larger context sizes.

* Fix test-backend-ops crash glu, get_rows, scale, rms_norm, add

* hardcoded name handling for rope_freqs.weight

* Suppress logging and add error handling to allow test-backend-ops to complete

* Fix MUL_MAT with broadcast; Add unsupported MUL_MAT FLASH_ATTN cases

* Use bias instead of zp in test-backend-ops

* Update OV in CI, Add OV CI Tests in GH Actions

* Temp fix for multithreading bug

* Update OV CI, fix review suggestions.

* fix editorconfig-checker, update docs

* Fix tabs to spaces for editorconfig-checker

* fix editorconfig-checker

* Update docs

* updated model link to be GGUF model links

* Remove GGML_CPU_REPACK=OFF

* Skip permuted ADD and MUL

* Removed static variables from utils.cpp

* Removed initializing non-existing variable

* Remove unused structs

* Fix test-backend-ops for OV GPU

* unify api calling

* Update utils.cpp

* When the dim is dynamic, throw an error, need to is stastic forst

* Add interface compute_model_outputs(), which get the model output through computing the node use count & status in the cgraph to avoid the flag using

* No need to return

* Fix test-backend-ops for OV GPU LNL

* Fix test-thread-safety

* use the shape from infer request of output tensor create to avoid issue

* fix dynamic output shape  issue

* fix issue for the unused node in tests

* Remove unused lock

* Add comment

* Update openvino docs

* update to OV release version 2026.0

* add ci ov-gpu self hosted runner

* fix editorconfig

* Fix perplexity

* Rewrite the model inputs finding mechanism  (#54)

* Rewrite the model inputs finding logistic

* Put stateful shape handle in get input shape

* Put the iteration logistic in func

* Added ggml-ci-intel-openvino-gpu and doc update

* .hpp files converted to .h

* fix ggml-ci-x64-intel-openvino-gpu

* Fix for stateful execution bug in llama-bench

* Minor updates after stateful llama-bench fix

* Update ggml/src/ggml-openvino/utils.cpp

Co-authored-by: Yamini Nimmagadda <yamini.nimmagadda@intel.com>

* Remove multiple get_shape calls

* Bring back mutex into compute

* Fix VIEW op, which slice the input node

* Added token_len_per_seq existence check before slicing masks and moved node retrieval inside guarded block to prevent missing-key access

* Temp. fix for test requant errors

* Update to OV ggml-ci to low-perf

* ci : temporary disable "test-llama-archs"

* ci : cache v4 -> v5, checkout v4 -> v6, fix runner tag

* docs : update url

* Fix OV link in docker and Update docs

---------

Co-authored-by: Ravi Panchumarthy <ravi.panchumarthy@intel.com>
Co-authored-by: Cavus Mustafa <mustafa.cavus@intel.com>
Co-authored-by: Arshath <arshath.ramzan@intel.com>
Co-authored-by: XuejunZhai <Xuejun.Zhai@intel.com>
Co-authored-by: Yamini Nimmagadda <yamini.nimmagadda@intel.com>
Co-authored-by: Xuejun Zhai <Xuejun.Zhai@intel>
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2026-03-14 07:56:55 +02:00

574 lines
17 KiB
C++

#include "ggml-backend-impl.h"
#include "ggml-backend.h"
#include "ggml-backend-dl.h"
#include "ggml-impl.h"
#include <algorithm>
#include <cstring>
#include <filesystem>
#include <memory>
#include <string>
#include <type_traits>
#include <vector>
#include <cctype>
#ifdef _WIN32
# define WIN32_LEAN_AND_MEAN
# ifndef NOMINMAX
# define NOMINMAX
# endif
# include <windows.h>
#elif defined(__APPLE__)
# include <mach-o/dyld.h>
# include <dlfcn.h>
#else
# include <dlfcn.h>
# include <unistd.h>
#endif
// Backend registry
#ifdef GGML_USE_CPU
#include "ggml-cpu.h"
#endif
#ifdef GGML_USE_CUDA
#include "ggml-cuda.h"
#endif
#ifdef GGML_USE_METAL
#include "ggml-metal.h"
#endif
#ifdef GGML_USE_SYCL
#include "ggml-sycl.h"
#endif
#ifdef GGML_USE_VULKAN
#include "ggml-vulkan.h"
#endif
#ifdef GGML_USE_WEBGPU
#include "ggml-webgpu.h"
#endif
#ifdef GGML_USE_ZDNN
#include "ggml-zdnn.h"
#endif
#ifdef GGML_USE_OPENCL
#include "ggml-opencl.h"
#endif
#ifdef GGML_USE_HEXAGON
#include "ggml-hexagon.h"
#endif
#ifdef GGML_USE_BLAS
#include "ggml-blas.h"
#endif
#ifdef GGML_USE_RPC
#include "ggml-rpc.h"
#endif
#ifdef GGML_USE_VIRTGPU_FRONTEND
#include "ggml-virtgpu.h"
#endif
#ifdef GGML_USE_CANN
#include "ggml-cann.h"
#endif
#ifdef GGML_USE_ZENDNN
#include "ggml-zendnn.h"
#endif
#ifdef GGML_USE_OPENVINO
#include "ggml-openvino.h"
#endif
namespace fs = std::filesystem;
static std::string path_str(const fs::path & path) {
try {
#if defined(__cpp_lib_char8_t)
// C++20 and later: u8string() returns std::u8string
const std::u8string u8str = path.u8string();
return std::string(reinterpret_cast<const char *>(u8str.data()), u8str.size());
#else
// C++17: u8string() returns std::string
return path.u8string();
#endif
} catch (...) {
return std::string();
}
}
struct ggml_backend_reg_entry {
ggml_backend_reg_t reg;
dl_handle_ptr handle;
};
struct ggml_backend_registry {
std::vector<ggml_backend_reg_entry> backends;
std::vector<ggml_backend_dev_t> devices;
ggml_backend_registry() {
#ifdef GGML_USE_CUDA
register_backend(ggml_backend_cuda_reg());
#endif
#ifdef GGML_USE_METAL
register_backend(ggml_backend_metal_reg());
#endif
#ifdef GGML_USE_SYCL
register_backend(ggml_backend_sycl_reg());
#endif
#ifdef GGML_USE_VULKAN
// Add runtime disable check
if (getenv("GGML_DISABLE_VULKAN") == nullptr) {
register_backend(ggml_backend_vk_reg());
} else {
GGML_LOG_DEBUG("Vulkan backend disabled by GGML_DISABLE_VULKAN environment variable\n");
}
#endif
#ifdef GGML_USE_WEBGPU
register_backend(ggml_backend_webgpu_reg());
#endif
#ifdef GGML_USE_ZDNN
register_backend(ggml_backend_zdnn_reg());
#endif
#ifdef GGML_USE_VIRTGPU_FRONTEND
register_backend(ggml_backend_virtgpu_reg());
#endif
#ifdef GGML_USE_OPENCL
register_backend(ggml_backend_opencl_reg());
#endif
#ifdef GGML_USE_ZENDNN
register_backend(ggml_backend_zendnn_reg());
#endif
#ifdef GGML_USE_HEXAGON
register_backend(ggml_backend_hexagon_reg());
#endif
#ifdef GGML_USE_CANN
register_backend(ggml_backend_cann_reg());
#endif
#ifdef GGML_USE_BLAS
register_backend(ggml_backend_blas_reg());
#endif
#ifdef GGML_USE_RPC
register_backend(ggml_backend_rpc_reg());
#endif
#ifdef GGML_USE_OPENVINO
register_backend(ggml_backend_openvino_reg());
#endif
#ifdef GGML_USE_CPU
register_backend(ggml_backend_cpu_reg());
#endif
}
~ggml_backend_registry() {
// FIXME: backends cannot be safely unloaded without a function to destroy all the backend resources,
// since backend threads may still be running and accessing resources from the dynamic library
for (auto & entry : backends) {
if (entry.handle) {
entry.handle.release(); // NOLINT
}
}
}
void register_backend(ggml_backend_reg_t reg, dl_handle_ptr handle = nullptr) {
if (!reg) {
return;
}
#ifndef NDEBUG
GGML_LOG_DEBUG("%s: registered backend %s (%zu devices)\n",
__func__, ggml_backend_reg_name(reg), ggml_backend_reg_dev_count(reg));
#endif
backends.push_back({ reg, std::move(handle) });
for (size_t i = 0; i < ggml_backend_reg_dev_count(reg); i++) {
register_device(ggml_backend_reg_dev_get(reg, i));
}
}
void register_device(ggml_backend_dev_t device) {
#ifndef NDEBUG
GGML_LOG_DEBUG("%s: registered device %s (%s)\n", __func__, ggml_backend_dev_name(device), ggml_backend_dev_description(device));
#endif
devices.push_back(device);
}
ggml_backend_reg_t load_backend(const fs::path & path, bool silent) {
dl_handle_ptr handle { dl_load_library(path) };
if (!handle) {
if (!silent) {
GGML_LOG_ERROR("%s: failed to load %s: %s\n", __func__, path_str(path).c_str(), dl_error());
}
return nullptr;
}
auto score_fn = (ggml_backend_score_t) dl_get_sym(handle.get(), "ggml_backend_score");
if (score_fn && score_fn() == 0) {
if (!silent) {
GGML_LOG_INFO("%s: backend %s is not supported on this system\n", __func__, path_str(path).c_str());
}
return nullptr;
}
auto backend_init_fn = (ggml_backend_init_t) dl_get_sym(handle.get(), "ggml_backend_init");
if (!backend_init_fn) {
if (!silent) {
GGML_LOG_ERROR("%s: failed to find ggml_backend_init in %s\n", __func__, path_str(path).c_str());
}
return nullptr;
}
ggml_backend_reg_t reg = backend_init_fn();
if (!reg || reg->api_version != GGML_BACKEND_API_VERSION) {
if (!silent) {
if (!reg) {
GGML_LOG_ERROR("%s: failed to initialize backend from %s: ggml_backend_init returned NULL\n",
__func__, path_str(path).c_str());
} else {
GGML_LOG_ERROR("%s: failed to initialize backend from %s: incompatible API version (backend: %d, current: %d)\n",
__func__, path_str(path).c_str(), reg->api_version, GGML_BACKEND_API_VERSION);
}
}
return nullptr;
}
GGML_LOG_INFO("%s: loaded %s backend from %s\n", __func__, ggml_backend_reg_name(reg), path_str(path).c_str());
register_backend(reg, std::move(handle));
return reg;
}
void unload_backend(ggml_backend_reg_t reg, bool silent) {
auto it = std::find_if(backends.begin(), backends.end(),
[reg](const ggml_backend_reg_entry & entry) { return entry.reg == reg; });
if (it == backends.end()) {
if (!silent) {
GGML_LOG_ERROR("%s: backend not found\n", __func__);
}
return;
}
if (!silent) {
GGML_LOG_DEBUG("%s: unloading %s backend\n", __func__, ggml_backend_reg_name(reg));
}
// remove devices
devices.erase(
std::remove_if(devices.begin(), devices.end(),
[reg](ggml_backend_dev_t dev) { return ggml_backend_dev_backend_reg(dev) == reg; }),
devices.end());
// remove backend
backends.erase(it);
}
};
static ggml_backend_registry & get_reg() {
static ggml_backend_registry reg;
return reg;
}
// Internal API
void ggml_backend_register(ggml_backend_reg_t reg) {
get_reg().register_backend(reg);
}
void ggml_backend_device_register(ggml_backend_dev_t device) {
get_reg().register_device(device);
}
// Backend (reg) enumeration
static bool striequals(const char * a, const char * b) {
for (; *a && *b; a++, b++) {
if (std::tolower(*a) != std::tolower(*b)) {
return false;
}
}
return *a == *b;
}
size_t ggml_backend_reg_count() {
return get_reg().backends.size();
}
ggml_backend_reg_t ggml_backend_reg_get(size_t index) {
GGML_ASSERT(index < ggml_backend_reg_count());
return get_reg().backends[index].reg;
}
ggml_backend_reg_t ggml_backend_reg_by_name(const char * name) {
for (size_t i = 0; i < ggml_backend_reg_count(); i++) {
ggml_backend_reg_t reg = ggml_backend_reg_get(i);
if (striequals(ggml_backend_reg_name(reg), name)) {
return reg;
}
}
return nullptr;
}
// Device enumeration
size_t ggml_backend_dev_count() {
return get_reg().devices.size();
}
ggml_backend_dev_t ggml_backend_dev_get(size_t index) {
GGML_ASSERT(index < ggml_backend_dev_count());
return get_reg().devices[index];
}
ggml_backend_dev_t ggml_backend_dev_by_name(const char * name) {
for (size_t i = 0; i < ggml_backend_dev_count(); i++) {
ggml_backend_dev_t dev = ggml_backend_dev_get(i);
if (striequals(ggml_backend_dev_name(dev), name)) {
return dev;
}
}
return nullptr;
}
ggml_backend_dev_t ggml_backend_dev_by_type(enum ggml_backend_dev_type type) {
for (size_t i = 0; i < ggml_backend_dev_count(); i++) {
ggml_backend_dev_t dev = ggml_backend_dev_get(i);
if (ggml_backend_dev_type(dev) == type) {
return dev;
}
}
return nullptr;
}
// Convenience functions
ggml_backend_t ggml_backend_init_by_name(const char * name, const char * params) {
ggml_backend_dev_t dev = ggml_backend_dev_by_name(name);
if (!dev) {
return nullptr;
}
return ggml_backend_dev_init(dev, params);
}
ggml_backend_t ggml_backend_init_by_type(enum ggml_backend_dev_type type, const char * params) {
ggml_backend_dev_t dev = ggml_backend_dev_by_type(type);
if (!dev) {
return nullptr;
}
return ggml_backend_dev_init(dev, params);
}
ggml_backend_t ggml_backend_init_best(void) {
ggml_backend_dev_t dev = ggml_backend_dev_by_type(GGML_BACKEND_DEVICE_TYPE_GPU);
dev = dev ? dev : ggml_backend_dev_by_type(GGML_BACKEND_DEVICE_TYPE_IGPU);
dev = dev ? dev : ggml_backend_dev_by_type(GGML_BACKEND_DEVICE_TYPE_CPU);
if (!dev) {
return nullptr;
}
return ggml_backend_dev_init(dev, nullptr);
}
// Dynamic loading
ggml_backend_reg_t ggml_backend_load(const char * path) {
return get_reg().load_backend(path, false);
}
void ggml_backend_unload(ggml_backend_reg_t reg) {
get_reg().unload_backend(reg, true);
}
static fs::path get_executable_path() {
#if defined(__APPLE__)
// get executable path
std::vector<char> path;
uint32_t size;
while (true) {
size = path.size();
if (_NSGetExecutablePath(path.data(), &size) == 0) {
break;
}
path.resize(size);
}
std::string base_path(path.data(), size);
// remove executable name
auto last_slash = base_path.find_last_of('/');
if (last_slash != std::string::npos) {
base_path = base_path.substr(0, last_slash);
}
return base_path + "/";
#elif defined(__linux__) || defined(__FreeBSD__)
std::string base_path = ".";
std::vector<char> path(1024);
while (true) {
// get executable path
# if defined(__linux__)
ssize_t len = readlink("/proc/self/exe", path.data(), path.size());
# elif defined(__FreeBSD__)
ssize_t len = readlink("/proc/curproc/file", path.data(), path.size());
# endif
if (len == -1) {
break;
}
if (len < (ssize_t) path.size()) {
base_path = std::string(path.data(), len);
// remove executable name
auto last_slash = base_path.find_last_of('/');
if (last_slash != std::string::npos) {
base_path = base_path.substr(0, last_slash);
}
break;
}
path.resize(path.size() * 2);
}
return base_path + "/";
#elif defined(_WIN32)
std::vector<wchar_t> path(MAX_PATH);
DWORD len = GetModuleFileNameW(NULL, path.data(), path.size());
if (len == 0) {
return {};
}
std::wstring base_path(path.data(), len);
// remove executable name
auto last_slash = base_path.find_last_of('\\');
if (last_slash != std::string::npos) {
base_path = base_path.substr(0, last_slash);
}
return base_path + L"\\";
#else
return {};
#endif
}
static fs::path backend_filename_prefix() {
#ifdef _WIN32
return fs::u8path("ggml-");
#else
return fs::u8path("libggml-");
#endif
}
static fs::path backend_filename_extension() {
#ifdef _WIN32
return fs::u8path(".dll");
#else
return fs::u8path(".so");
#endif
}
static ggml_backend_reg_t ggml_backend_load_best(const char * name, bool silent, const char * user_search_path) {
// enumerate all the files that match [lib]ggml-name-*.[so|dll] in the search paths
const fs::path name_path = fs::u8path(name);
const fs::path file_prefix = backend_filename_prefix().native() + name_path.native() + fs::u8path("-").native();
const fs::path file_extension = backend_filename_extension();
std::vector<fs::path> search_paths;
if (user_search_path == nullptr) {
#ifdef GGML_BACKEND_DIR
search_paths.push_back(fs::u8path(GGML_BACKEND_DIR));
#endif
// default search paths: executable directory, current directory
search_paths.push_back(get_executable_path());
search_paths.push_back(fs::current_path());
} else {
search_paths.push_back(fs::u8path(user_search_path));
}
int best_score = 0;
fs::path best_path;
std::error_code ec;
for (const auto & search_path : search_paths) {
if (!fs::exists(search_path, ec)) {
if (ec) {
GGML_LOG_DEBUG("%s: posix_stat(%s) failure, error-message: %s\n", __func__, path_str(search_path).c_str(), ec.message().c_str());
} else {
GGML_LOG_DEBUG("%s: search path %s does not exist\n", __func__, path_str(search_path).c_str());
}
continue;
}
fs::directory_iterator dir_it(search_path, fs::directory_options::skip_permission_denied);
for (const auto & entry : dir_it) {
if (entry.is_regular_file(ec)) {
auto filename = entry.path().filename();
auto ext = entry.path().extension();
if (filename.native().find(file_prefix) == 0 && ext == file_extension) {
dl_handle_ptr handle { dl_load_library(entry) };
if (!handle && !silent) {
GGML_LOG_ERROR("%s: failed to load %s: %s\n", __func__, path_str(entry.path()).c_str(), dl_error());
}
if (handle) {
auto score_fn = (ggml_backend_score_t) dl_get_sym(handle.get(), "ggml_backend_score");
if (score_fn) {
int s = score_fn();
#ifndef NDEBUG
GGML_LOG_DEBUG("%s: %s score: %d\n", __func__, path_str(entry.path()).c_str(), s);
#endif
if (s > best_score) {
best_score = s;
best_path = entry.path();
}
} else {
if (!silent) {
GGML_LOG_INFO("%s: failed to find ggml_backend_score in %s\n", __func__, path_str(entry.path()).c_str());
}
}
}
}
}
}
}
if (best_score == 0) {
// try to load the base backend
for (const auto & search_path : search_paths) {
fs::path filename = backend_filename_prefix().native() + name_path.native() + backend_filename_extension().native();
fs::path path = search_path / filename;
if (std::error_code ec; fs::exists(path, ec)) {
return get_reg().load_backend(path, silent);
} else {
if (ec) {
GGML_LOG_DEBUG("%s: posix_stat(%s) failure, error-message: %s\n", __func__, path_str(path).c_str(), ec.message().c_str());
}
}
}
return nullptr;
}
return get_reg().load_backend(best_path, silent);
}
void ggml_backend_load_all() {
ggml_backend_load_all_from_path(nullptr);
}
void ggml_backend_load_all_from_path(const char * dir_path) {
#ifdef NDEBUG
bool silent = true;
#else
bool silent = false;
#endif
ggml_backend_load_best("blas", silent, dir_path);
ggml_backend_load_best("zendnn", silent, dir_path);
ggml_backend_load_best("cann", silent, dir_path);
ggml_backend_load_best("cuda", silent, dir_path);
ggml_backend_load_best("hip", silent, dir_path);
ggml_backend_load_best("metal", silent, dir_path);
ggml_backend_load_best("rpc", silent, dir_path);
ggml_backend_load_best("sycl", silent, dir_path);
ggml_backend_load_best("vulkan", silent, dir_path);
ggml_backend_load_best("virtgpu", silent, dir_path);
ggml_backend_load_best("opencl", silent, dir_path);
ggml_backend_load_best("hexagon", silent, dir_path);
ggml_backend_load_best("musa", silent, dir_path);
ggml_backend_load_best("openvino", silent, dir_path);
ggml_backend_load_best("cpu", silent, dir_path);
// check the environment variable GGML_BACKEND_PATH to load an out-of-tree backend
const char * backend_path = std::getenv("GGML_BACKEND_PATH");
if (backend_path) {
ggml_backend_load(backend_path);
}
}