Skip to content

parallel : increase the variability of the prompt lengths #13927

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
May 30, 2025
Merged

parallel : increase the variability of the prompt lengths #13927

Show file tree
Hide file tree
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
2 changes: 1 addition & 1 deletion examples/parallel/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,7 +4,7 @@ Simplified simulation of serving incoming requests in parallel

## Example

Generate 128 client requests (`-ns 128`), simulating 8 concurrent clients (`-np 8`). The system prompt is shared (`-pps`), meaning that it is computed once at the start. The client requests consist of 10 junk questions (`-j 10`) followed by the actual question.
Generate 128 client requests (`-ns 128`), simulating 8 concurrent clients (`-np 8`). The system prompt is shared (`-pps`), meaning that it is computed once at the start. The client requests consist of up to 10 junk questions (`--junk 10`) followed by the actual question.

```bash
llama-parallel -m model.gguf -np 8 -ns 128 --top-k 1 -pps --junk 10 -c 16384
Expand Down
7 changes: 5 additions & 2 deletions examples/parallel/parallel.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -315,7 +315,10 @@ int main(int argc, char ** argv) {
} else {
client.prompt += k_system;
}
for (int i = 0; i < n_junk; ++i) {

const int n_junk_cur = rand() % n_junk;
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

When n_junk is zero this is a division by zero.

Suggested change
const int n_junk_cur = rand() % n_junk;
const int n_junk_cur = n_junk > 0 ? rand() % n_junk : 0;

Since n_junk = 0 is the default, this throws a floating point exception by default.

llama.cpp/examples/parallel/parallel.cpp

Line 161 in bc6e86b

params.n_junk = 0;


for (int i = 0; i < n_junk_cur; ++i) {
const int r = rand() % k_questions.size();
client.prompt += "User:\n" + k_questions[r] + "\nAssistant:\n " + k_answers[r] + "\n";
}
Expand All @@ -340,7 +343,7 @@ int main(int argc, char ** argv) {
client.n_decoded = 0;
client.i_batch = batch.n_tokens - 1;

LOG_INF("\033[31mClient %3d, seq %4d, started decoding ...\033[0m\n", client.id, client.seq_id);
LOG_INF("\033[31mClient %3d, seq %4d, junk = %4d, started decoding ...\033[0m\n", client.id, client.seq_id, n_junk_cur);

g_seq_id += 1;

Expand Down
Loading