happyz
/
llama.cpp
mirror of https://github.com/ggerganov/llama.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

metal : adaptive CPU/GPU interleave based on number of nodes (#19369)

This commit is contained in:
Georgi Gerganov 2026-02-05 19:07:22 +02:00 committed by GitHub
parent 449ec2ab07
commit 22cae83218
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
ggml/src/ggml-metal/ggml-metal-context.m
View File

@ -415,7 +415,7 @@ bool ggml_metal_cpy_tensor_async(ggml_metal_t ctx_src, ggml_metal_t ctx_dst, con
enum ggml_status ggml_metal_graph_compute(ggml_metal_t ctx, struct ggml_cgraph * gf) { enum ggml_status ggml_metal_graph_compute(ggml_metal_t ctx, struct ggml_cgraph * gf) {
// number of nodes encoded by the main thread (empirically determined) // number of nodes encoded by the main thread (empirically determined)
const int n_main = 64; const int n_main = MAX(64, 0.1*gf->n_nodes);
// number of threads in addition to the main thread // number of threads in addition to the main thread
const int n_cb = ctx->n_cb; const int n_cb = ctx->n_cb;