ollama/llama/ggml-metal.h

/**
 * llama.cpp - commit 46e3556e01b824e52395fb050b29804b6cff2a7c - do not edit this file
 *
 * MIT License
 *
 * Copyright (c) 2023-2024 The ggml authors
 *
 * Permission is hereby granted, free of charge, to any person obtaining a copy
 * of this software and associated documentation files (the "Software"), to deal
 * in the Software without restriction, including without limitation the rights
 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 * copies of the Software, and to permit persons to whom the Software is
 * furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in all
 * copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 * SOFTWARE.
 */

// Note: this description is outdated
//
// An interface allowing to compute ggml_cgraph with Metal
//
// This is a fully functional interface that extends ggml with GPU support for Apple devices.
// A similar interface can be created for other GPU backends (e.g. Vulkan, CUDA, etc.)
//
// How it works?
//
// As long as your program can create and evaluate a ggml_cgraph on the CPU, you can use this
// interface to evaluate the same graph on the GPU. Instead of using ggml_graph_compute(), you
// use ggml_metal_graph_compute() (or ggml_vulkan_graph_compute(), etc.)
//
// You only need to make sure that all memory buffers that you used during the graph creation
// are mapped to the device memory with the ggml_metal_add_buffer() function. This mapping is
// used during the graph evaluation to determine the arguments of the compute kernels.
//
// Synchronization between device and host memory (for example for input and output tensors)
// is done with the ggml_metal_set_tensor() and ggml_metal_get_tensor() functions.
//

#pragma once

#include "ggml.h"
#include "ggml-backend.h"

#include <stddef.h>
#include <stdbool.h>

struct ggml_tensor;
struct ggml_cgraph;

#ifdef __cplusplus
extern "C" {
#endif

//
// backend API
// user-code should use only these functions
//

GGML_BACKEND_API ggml_backend_t ggml_backend_metal_init(void);

GGML_BACKEND_API bool ggml_backend_is_metal(ggml_backend_t backend);

GGML_DEPRECATED(
        GGML_BACKEND_API ggml_backend_buffer_t ggml_backend_metal_buffer_from_ptr(void * data, size_t size, size_t max_size),
        "obsoleted by the new device interface - https://github.com/ggerganov/llama.cpp/pull/9713");

GGML_BACKEND_API void ggml_backend_metal_set_abort_callback(ggml_backend_t backend, ggml_abort_callback abort_callback, void * user_data);

GGML_BACKEND_API ggml_backend_buffer_type_t ggml_backend_metal_buffer_type(void);

// helper to check if the device supports a specific family
// ideally, the user code should be doing these checks
// ref: https://developer.apple.com/metal/Metal-Feature-Set-Tables.pdf
GGML_BACKEND_API bool ggml_backend_metal_supports_family(ggml_backend_t backend, int family);

// capture all command buffers committed the next time `ggml_backend_graph_compute` is called
GGML_BACKEND_API void ggml_backend_metal_capture_next_compute(ggml_backend_t backend);

GGML_BACKEND_API ggml_backend_reg_t ggml_backend_metal_reg(void);

#ifdef __cplusplus
}
#endif