Skip to content

llama-bench : add --no-warmup flag (#14224) #14270

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Jun 19, 2025
Merged

llama-bench : add --no-warmup flag (#14224) #14270

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
44 changes: 26 additions & 18 deletions tools/llama-bench/llama-bench.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -267,6 +267,7 @@ struct cmd_params {
int delay;
bool verbose;
bool progress;
bool no_warmup;
output_formats output_format;
output_formats output_format_stderr;
};
Expand Down Expand Up @@ -303,6 +304,7 @@ static const cmd_params cmd_params_defaults = {
/* delay */ 0,
/* verbose */ false,
/* progress */ false,
/* no_warmup */ false,
/* output_format */ MARKDOWN,
/* output_format_stderr */ NONE,
};
Expand All @@ -325,6 +327,7 @@ static void print_usage(int /* argc */, char ** argv) {
output_format_str(cmd_params_defaults.output_format_stderr));
printf(" -v, --verbose verbose output\n");
printf(" --progress print test progress indicators\n");
printf(" --no-warmup skip warmup runs before benchmarking\n");
printf("\n");
printf("test parameters:\n");
printf(" -m, --model <filename> (default: %s)\n", join(cmd_params_defaults.model, ",").c_str());
Expand Down Expand Up @@ -425,6 +428,7 @@ static cmd_params parse_cmd_params(int argc, char ** argv) {
params.prio = cmd_params_defaults.prio;
params.delay = cmd_params_defaults.delay;
params.progress = cmd_params_defaults.progress;
params.no_warmup = cmd_params_defaults.no_warmup;

for (int i = 1; i < argc; i++) {
arg = argv[i];
Expand Down Expand Up @@ -798,6 +802,8 @@ static cmd_params parse_cmd_params(int argc, char ** argv) {
params.verbose = true;
} else if (arg == "--progress") {
params.progress = true;
} else if (arg == "--no-warmup") {
params.no_warmup = true;
} else {
invalid_param = true;
break;
Expand Down Expand Up @@ -1925,25 +1931,27 @@ int main(int argc, char ** argv) {
llama_attach_threadpool(ctx, threadpool, NULL);

// warmup run
if (t.n_prompt > 0) {
if (params.progress) {
fprintf(stderr, "llama-bench: benchmark %d/%zu: warmup prompt run\n", params_idx, params_count);
}
//test_prompt(ctx, std::min(t.n_batch, std::min(t.n_prompt, 32)), 0, t.n_batch, t.n_threads);
bool res = test_prompt(ctx, t.n_prompt, t.n_batch, t.n_threads);
if (!res) {
fprintf(stderr, "%s: error: failed to run prompt warmup\n", __func__);
exit(1);
}
}
if (t.n_gen > 0) {
if (params.progress) {
fprintf(stderr, "llama-bench: benchmark %d/%zu: warmup generation run\n", params_idx, params_count);
if (!params.no_warmup) {
if (t.n_prompt > 0) {
if (params.progress) {
fprintf(stderr, "llama-bench: benchmark %d/%zu: warmup prompt run\n", params_idx, params_count);
}
//test_prompt(ctx, std::min(t.n_batch, std::min(t.n_prompt, 32)), 0, t.n_batch, t.n_threads);
bool res = test_prompt(ctx, t.n_prompt, t.n_batch, t.n_threads);
if (!res) {
fprintf(stderr, "%s: error: failed to run prompt warmup\n", __func__);
exit(1);
}
}
bool res = test_gen(ctx, 1, t.n_threads);
if (!res) {
fprintf(stderr, "%s: error: failed to run gen warmup\n", __func__);
exit(1);
if (t.n_gen > 0) {
if (params.progress) {
fprintf(stderr, "llama-bench: benchmark %d/%zu: warmup generation run\n", params_idx, params_count);
}
bool res = test_gen(ctx, 1, t.n_threads);
if (!res) {
fprintf(stderr, "%s: error: failed to run gen warmup\n", __func__);
exit(1);
}
}
}

Expand Down
Loading