ikawrakow/ik_llama.cpp
1
0
Fork 0
mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-04-24 16:39:45 +00:00
Code Issues Packages Projects Releases Wiki Activity

Be able to set a max. number of GPUs to be used in split mode graph (#1051)

Browse Source 
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
This commit is contained in:
Kawrakow
2025-12-11 07:22:53 +01:00
committed by GitHub
parent 6a5a707ac0
commit 9484d150d8
6 changed files with 67 additions and 19 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
common/common.h
View File

@@ -154,6 +154,7 @@ struct gpt_params {
int32_t n_gpu_layers = -1; // number of layers to store in VRAM (-1 - use default)
int32_t n_gpu_layers_draft = -1; // number of layers to store in VRAM for the draft model (-1 - use default)
int32_t main_gpu = 0; // the GPU that is used for scratch and small tensors
int32_t max_gpu = 0; // max number of GPUs to use at a time for split mode "graph"
float tensor_split[128] = {0}; // how split tensors should be distributed across GPUs
int32_t grp_attn_n = 1; // group-attention factor
int32_t grp_attn_w = 512; // group-attention width