vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

llama : sync with recent PRs on master

Browse source
This commit is contained in:
Georgi Gerganov 2023-08-17 19:03:15 +03:00
parent 8ace03ad3d
commit 11bf4366c2
No known key found for this signature in database
GPG key ID: 449E073F9DC10735
7 changed files with 616 additions and 672 deletions
Download patch file Download diff file Expand all files Collapse all files

9
ggml-metal.h
View file

@ -66,10 +66,13 @@ void ggml_metal_get_tensor(struct ggml_metal_context * ctx, struct ggml_tensor *
// 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);
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
bool ggml_metal_if_optimized(struct ggml_metal_context * ctx);
// 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