vrr/prima.cpp
2
0
Fork 0
mirror of https://github.com/Lizonghang/prima.cpp.git synced 2025-09-11 06:04:50 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

fix flops count and ram/vram speed test

Browse source
This commit is contained in:
Zonghang Li 2024-12-08 10:14:05 +04:00
parent 26c2ffb5b7
commit df813675d0
5 changed files with 136 additions and 83 deletions
Download patch file Download diff file Expand all files Collapse all files

3
common/common.cpp
View file

@ -896,7 +896,7 @@ struct llama_init_result llama_init_from_gpt_params(gpt_params & params) {
device_info dev_info;
dev_info.rank = params.rank;
llama_profile_device(&dev_info, model, ml, params.n_predict, params.cpuparams.n_threads);
llama_profile_device(&dev_info, model, ml, params.n_predict, params.n_ctx, params.cpuparams.n_threads, params.flash_attn);
// create llama context
struct llama_context_params cparams = llama_context_params_from_gpt_params(params);
@ -1133,6 +1133,7 @@ struct llama_context_params llama_context_params_from_gpt_params(const gpt_param
std::strcpy(cparams.next_node_ip, params.next_node_ip.c_str());
cparams.n_ctx = params.n_ctx;
cparams.n_predict = params.n_predict;
cparams.n_seq_max = params.n_parallel;
cparams.n_batch = params.n_batch;
cparams.n_ubatch = params.n_ubatch;