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

metal : enable ggml-alloc (#2627)

Browse source 
* metal: enable ggml-alloc

Make ggml-alloc work with concurrently dispatch.

* style-fix

Co-authored-by: slaren <slarengh@gmail.com>

---------

Co-authored-by: slaren <slarengh@gmail.com>
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
This commit is contained in:
Shouzheng Liu 2023-08-16 16:08:28 -04:00 • committed by GitHub
parent bf83bff674
commit fc8ef549e5
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
5 changed files with 61 additions and 26 deletions
Download patch file Download diff file Expand all files Collapse all files

9
ggml-metal.h
View file

@ -63,10 +63,13 @@ 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