llama.cpp/ggml-kompute.h

70 lines
1.7 KiB
C
Raw Normal View History

#pragma once
#include "ggml-backend.h"
#include <cstddef>
#include <vector>
#include <string>
struct ggml_kompute_context;
namespace vk {
class DeviceMemory;
class Buffer;
};
struct ggml_vk_memory {
void *data = nullptr;
size_t size = 0;
vk::DeviceMemory *primaryMemory = nullptr;
vk::Buffer *primaryBuffer = nullptr;
vk::DeviceMemory *stagingMemory = nullptr;
vk::Buffer *stagingBuffer = nullptr;
};
struct ggml_vk_device {
int index = 0;
int type = 0; // same as VkPhysicalDeviceType
size_t heapSize = 0;
std::string name;
std::string vendor;
int subgroupSize = 0;
};
std::vector<ggml_vk_device> ggml_vk_available_devices(size_t memoryRequired);
bool ggml_vk_init_device(size_t memoryRequired, const std::string &device);
bool ggml_vk_init_device(const ggml_vk_device &device);
bool ggml_vk_init_device(int device);
bool ggml_vk_free_device();
bool ggml_vk_has_vulkan();
bool ggml_vk_has_device();
bool ggml_vk_using_vulkan();
ggml_vk_device ggml_vk_current_device();
struct ggml_kompute_context * ggml_vk_init(void);
void ggml_vk_free(struct ggml_kompute_context * ctx);
void ggml_vk_free_memory(ggml_vk_memory &memory);
void ggml_vk_graph_compute(struct ggml_kompute_context * ctx, struct ggml_cgraph * gf);
//
// backend API
// user-code should use only these functions
//
#ifdef __cplusplus
extern "C" {
#endif
// forward declaration
typedef struct ggml_backend * ggml_backend_t;
GGML_API ggml_backend_t ggml_backend_kompute_init(void);
GGML_API bool ggml_backend_is_kompute(ggml_backend_t backend);
GGML_API ggml_backend_buffer_type_t ggml_backend_kompute_buffer_type(void);
#ifdef __cplusplus
}
#endif