mirror of
https://github.com/LostRuins/koboldcpp.git
synced 2025-09-13 10:29:43 +00:00
e692a79aab
# Conflicts: # .github/workflows/docker.yml # CMakeLists.txt # CONTRIBUTING.md # docs/android.md # docs/docker.md # examples/embedding/embedding.cpp # examples/imatrix/imatrix.cpp # examples/infill/infill.cpp # examples/llama-bench/llama-bench.cpp # examples/main/README.md # examples/parallel/parallel.cpp # examples/perplexity/perplexity.cpp # examples/quantize-stats/quantize-stats.cpp # examples/save-load-state/save-load-state.cpp # examples/server/README.md # examples/simple/CMakeLists.txt # examples/speculative/speculative.cpp # flake.lock # ggml/src/CMakeLists.txt # ggml/src/ggml-blas.cpp # pocs/vdot/q8dot.cpp # pocs/vdot/vdot.cpp # scripts/debug-test.sh # scripts/sync-ggml.last # src/llama.cpp # tests/test-backend-ops.cpp # tests/test-chat-template.cpp # tests/test-quantize-fns.cpp # tests/test-quantize-perf.cpp # tests/test-tokenizer-0.cpp # tests/test-tokenizer-1-bpe.cpp # tests/test-tokenizer-1-spm.cpp
66 lines
2.3 KiB
C
66 lines
2.3 KiB
C
// Note: this description is outdated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, OpenCL, 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_API ggml_backend_t ggml_backend_metal_init(void);
|
|
|
|
GGML_API bool ggml_backend_is_metal(ggml_backend_t backend);
|
|
|
|
GGML_DEPRECATED(
|
|
GGML_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_API void ggml_backend_metal_set_abort_callback(ggml_backend_t backend, ggml_abort_callback abort_callback, void * user_data);
|
|
|
|
GGML_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_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_API void ggml_backend_metal_capture_next_compute(ggml_backend_t backend);
|
|
|
|
GGML_API ggml_backend_reg_t ggml_backend_metal_reg(void);
|
|
|
|
#ifdef __cplusplus
|
|
}
|
|
#endif
|