mirror of
https://github.com/ggerganov/llama.cpp.git
synced 2024-12-27 22:59:24 +01:00
f010b77a37
* vulkan : add backend registry / device interfaces * llama : print devices used on model load
32 lines
932 B
C
32 lines
932 B
C
#pragma once
|
|
|
|
#include "ggml.h"
|
|
#include "ggml-backend.h"
|
|
|
|
#ifdef __cplusplus
|
|
extern "C" {
|
|
#endif
|
|
|
|
#define GGML_VK_NAME "Vulkan"
|
|
#define GGML_VK_MAX_DEVICES 16
|
|
|
|
GGML_API void ggml_vk_instance_init(void);
|
|
|
|
// backend API
|
|
GGML_API ggml_backend_t ggml_backend_vk_init(size_t dev_num);
|
|
|
|
GGML_API bool ggml_backend_is_vk(ggml_backend_t backend);
|
|
GGML_API int ggml_backend_vk_get_device_count(void);
|
|
GGML_API void ggml_backend_vk_get_device_description(int device, char * description, size_t description_size);
|
|
GGML_API void ggml_backend_vk_get_device_memory(int device, size_t * free, size_t * total);
|
|
|
|
GGML_API ggml_backend_buffer_type_t ggml_backend_vk_buffer_type(size_t dev_num);
|
|
// 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_vk_host_buffer_type(void);
|
|
|
|
GGML_API ggml_backend_reg_t ggml_backend_vk_reg(void);
|
|
|
|
#ifdef __cplusplus
|
|
}
|
|
#endif
|