vrr/prima.cpp
2
0
Fork 0
mirror of https://github.com/Lizonghang/prima.cpp.git synced 2025-09-06 15:59:07 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

fix t_load_us

Browse source
This commit is contained in:
Lizonghang 2024-11-28 15:55:21 +04:00
parent 740f7f0b95
commit 9a7bbce7ad
3 changed files with 8 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
common/common.cpp
View file

@ -944,6 +944,8 @@ struct llama_init_result llama_init_from_gpt_params(gpt_params & params) {
return iparams;
}
llama_perf_context_sync(lctx, model);
if (llama_context_setup_backend(model, cparams, lctx) == nullptr) {
LOG_ERR("%s: failed to setup context with model '%s'\n", __func__, params.model.c_str());
llama_free_model(model);

1
include/llama.h
View file

@ -1255,6 +1255,7 @@ extern "C" {
LLAMA_API struct llama_perf_context_data llama_perf_context (const struct llama_context * ctx);
LLAMA_API void llama_perf_context_print(const struct llama_context * ctx);
LLAMA_API void llama_perf_context_reset( struct llama_context * ctx);
LLAMA_API void llama_perf_context_sync ( struct llama_context * ctx, const struct llama_model * model);
// NOTE: the following work only with samplers constructed via llama_sampler_chain_init
LLAMA_API struct llama_perf_sampler_data llama_perf_sampler (const struct llama_sampler * chain);

5
src/llama.cpp
View file

@ -3547,6 +3547,11 @@ static ggml_backend_buffer_type_t llama_default_buffer_type_offload(const llama_
GGML_UNUSED(model);
}
void llama_perf_context_sync(struct llama_context * ctx, const struct llama_model * model) {
ctx->t_start_us = model->t_start_us;
ctx->t_load_us = model->t_load_us;
}
void llama_profile_device(device_info * dev_info, struct llama_model * model, llama_model_loader * ml, int n_threads) {
dev_info->device_name = device_name();
dev_info->cpu_props.cores = device_cpu_cores();