sleepy/llama.cpp
1
0
Fork 0
Code Issues 16 Pull Requests Actions 69 Packages Projects Releases Wiki Activity

common : more accurate sampling timing (#17382)

Browse Source 
* common : more accurate sampling timing

* eval-callback : minor fixes

* cont : add time_meas impl

* cont : fix log msg [no ci]

* cont : fix multiple definitions of time_meas

* llama-cli : exclude chat template init from time measurement

* cont : print percentage of unaccounted time

* cont : do not reset timings
This commit is contained in:
Georgi Gerganov
2025-11-20 13:40:10 +02:00
committed by GitHub
parent 5088b435d4
commit 196f5083ef
7 changed files with 102 additions and 33 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/llama-impl.cpp
+3 -3
View File
@@ -20,10 +20,10 @@ static llama_logger_state g_logger_state;
time_meas::time_meas(int64_t & t_acc, bool disable) : t_start_us(disable ? -1 : ggml_time_us()), t_acc(t_acc) {}
time_meas::~time_meas() {
if (t_start_us >= 0) {
t_acc += ggml_time_us() - t_start_us;
}
if (t_start_us >= 0) {
t_acc += ggml_time_us() - t_start_us;
}
}
void llama_log_set(ggml_log_callback log_callback, void * user_data) {
ggml_log_set(log_callback, user_data);