vrr/prima.cpp
2
0
Fork 0
mirror of https://github.com/Lizonghang/prima.cpp.git synced 2025-09-10 04:34:34 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

add arg --cuda-mem

Browse source
This commit is contained in:
Zonghang Li 2025-01-16 09:15:34 +04:00
parent dab6b2e1c2
commit 46e99218b4
5 changed files with 19 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

3
src/llama.cpp
View file

@ -3574,6 +3574,7 @@ void llama_profile_device(
device_info * dev_info,
struct llama_model * model,
llama_model_loader * ml,
int cuda_mem,
int n_predict,
int n_ctx,
int n_threads,
@ -3619,7 +3620,7 @@ void llama_profile_device(
dev_info->gpu_props.name = gpu_props.name;
dev_info->gpu_props.description = gpu_props.description;
dev_info->gpu_props.memory_free = round(gpu_props.memory_free / (double)(1 << 30) * 100) / 100;
dev_info->gpu_props.memory_free = std::min((double)cuda_mem, round(gpu_props.memory_free / (double)(1 << 30) * 100) / 100);
dev_info->gpu_props.memory_total = round(gpu_props.memory_total / (double)(1 << 30) * 100) / 100;
dev_info->gpu_props.metal_read_vram_bw = device_metal_read_vram_bw();
dev_info->gpu_props.cuda_read_vram_bw = device_cuda_read_vram_bw();