mirror of
https://github.com/ggerganov/llama.cpp.git
synced 2024-11-11 21:39:52 +00:00
e7e4df031b
* llama : ggml-backend integration * ggml-backend : add names to buffers * fix unmap after loading * batched-bench : add tensor_split param * llama : check for null tensor_split * ggml-backend : increase GGML_MAX_BACKENDS * improve graph splitting, partial fix for --no-kv-offload * cuda : add ggml-backend split buffer support * cuda : do not create buffer types for devices that don't exist (fixes usage without CUDA devices available) * ggml : fix null backend dereference (#4807) * ggml : fix null backend dereference * ggml : also check ggml_backend_is_cpu * test-backend-ops : check buffer allocation failures * llama : add cparam (split_mode) and command line argument (--split-mode, -sm) to configure the split mode (none, layer or row) * ggml : fix mul_mat_id work size * llama : rewrite session kv load/set without graphs * minor * llama : only initialize used backends, free backends on context free * llama : abort ctx if cuda backend init fails * llama : rewrite lora with ggml-backend and compute on CPU ggml-ci * llama : only map to a backend buffer the region of the file mapping containing the tensors used in the buffer * opencl : add ggml-backend buffer type * cuda : only use batched_cublas with batched mat muls (fixes fp16 tg perf) * llama : on Metal, by default offload the full model ggml-ci * metal : page align the data ptr (#4854) * Apply suggestions from code review Co-authored-by: Johannes Gäßler <johannesg@5d6.de> * cuda : fix split buffer free * address review comments * llama-bench : add split-mode parameter * fix whitespace * opencl : fix double initialization * server : add --split-mode parameter * use async copy and compute to improve multi-gpu performance ggml-ci * use async memcpys to copy the graph outputs to the CPU * fix opencl * use a host buffer for the cpu compute buffer for faster copies to the gpu --------- Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
53 lines
1.9 KiB
C
53 lines
1.9 KiB
C
#pragma once
|
|
|
|
#include "ggml.h"
|
|
#include "ggml-backend.h"
|
|
|
|
#ifdef GGML_USE_HIPBLAS
|
|
#define GGML_CUDA_NAME "ROCm"
|
|
#define GGML_CUBLAS_NAME "hipBLAS"
|
|
#else
|
|
#define GGML_CUDA_NAME "CUDA"
|
|
#define GGML_CUBLAS_NAME "cuBLAS"
|
|
#endif
|
|
|
|
#ifdef __cplusplus
|
|
extern "C" {
|
|
#endif
|
|
|
|
#define GGML_CUDA_MAX_DEVICES 16
|
|
|
|
// Always success. To check if CUDA is actually loaded, use `ggml_cublas_loaded`.
|
|
GGML_API void ggml_init_cublas(void);
|
|
|
|
// Returns `true` if there are available CUDA devices and cublas loads successfully; otherwise, it returns `false`.
|
|
GGML_API bool ggml_cublas_loaded(void);
|
|
|
|
GGML_API void * ggml_cuda_host_malloc(size_t size);
|
|
GGML_API void ggml_cuda_host_free(void * ptr);
|
|
|
|
GGML_API bool ggml_cuda_can_mul_mat(const struct ggml_tensor * src0, const struct ggml_tensor * src1, struct ggml_tensor * dst);
|
|
GGML_API bool ggml_cuda_compute_forward(struct ggml_compute_params * params, struct ggml_tensor * tensor);
|
|
|
|
GGML_API int ggml_cuda_get_device_count(void);
|
|
GGML_API void ggml_cuda_get_device_description(int device, char * description, size_t description_size);
|
|
|
|
// backend API
|
|
GGML_API ggml_backend_t ggml_backend_cuda_init(int device);
|
|
|
|
GGML_API bool ggml_backend_is_cuda(ggml_backend_t backend);
|
|
|
|
GGML_API ggml_backend_buffer_type_t ggml_backend_cuda_buffer_type(int device);
|
|
// split tensor buffer that splits matrices by rows across multiple devices
|
|
GGML_API ggml_backend_buffer_type_t ggml_backend_cuda_split_buffer_type(const float * tensor_split);
|
|
// pinned host buffer for use with the CPU backend for faster copies between CPU and GPU
|
|
GGML_API ggml_backend_buffer_type_t ggml_backend_cuda_host_buffer_type(void);
|
|
|
|
GGML_API int ggml_backend_cuda_get_device_count(void);
|
|
GGML_API void ggml_backend_cuda_get_device_description(int device, char * description, size_t description_size);
|
|
GGML_API void ggml_backend_cuda_get_device_memory(int device, size_t * free, size_t * total);
|
|
|
|
#ifdef __cplusplus
|
|
}
|
|
#endif
|