alihan/whisper.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/whisper.cpp.git synced 2023-11-04 02:52:44 +03:00
Code Issues Packages Projects Releases Wiki Activity

ggml : sync (ggml-alloc, GPU, eps, etc.) (#1220)

Browse Source 
* ggml : sync (ggml-alloc, GPU, eps, etc.)

* ggml : fix build

* wasm : fix build
This commit is contained in:
Georgi Gerganov
2023-09-05 13:54:40 +03:00
committed by GitHub
parent 6780c98e19
commit 59a3d0cb57
17 changed files with 11254 additions and 4095 deletions
Download Patch File Download Diff File Expand all files Collapse all files

20
ggml-metal.h
View File

@@ -24,6 +24,7 @@
// max memory buffers that can be mapped to the device
#define GGML_METAL_MAX_BUFFERS 16
#define GGML_METAL_MAX_COMMAND_BUFFERS 32
struct ggml_tensor;
struct ggml_cgraph;
@@ -34,9 +35,16 @@ extern "C" {
struct ggml_metal_context;
struct ggml_metal_context * ggml_metal_init(void);
// number of command buffers to use
struct ggml_metal_context * ggml_metal_init(int n_cb);
void ggml_metal_free(struct ggml_metal_context * ctx);
void * ggml_metal_host_malloc(size_t n);
void ggml_metal_host_free (void * data);
// set the number of command buffers to use
void ggml_metal_set_n_cb(struct ggml_metal_context * ctx, int n_cb);
// creates a mapping between a host memory buffer and a device memory buffer
// - make sure to map all buffers used in the graph before calling ggml_metal_graph_compute
// - the mapping is used during computation to determine the arguments of the compute kernels
@@ -57,6 +65,16 @@ void ggml_metal_set_tensor(struct ggml_metal_context * ctx, struct ggml_tensor *
// get data from the device into host memory
void ggml_metal_get_tensor(struct ggml_metal_context * ctx, struct ggml_tensor * t);
// try to find operations that can be run concurrently in the graph
// you should run it again if the topology of your graph changes
void ggml_metal_graph_find_concurrency(struct ggml_metal_context * ctx, struct ggml_cgraph * gf, bool check_mem);
// if the graph has been optimized for concurrently dispatch, return length of the concur_list if optimized
int ggml_metal_if_optimized(struct ggml_metal_context * ctx);
// output the concur_list for ggml_alloc
int * ggml_metal_get_concur_list(struct ggml_metal_context * ctx);
// same as ggml_graph_compute but uses Metal
// creates gf->n_threads command buffers in parallel
void ggml_metal_graph_compute(struct ggml_metal_context * ctx, struct ggml_cgraph * gf);