---
title: "xcsp3-2025-parallel"
description: "The XCSP3 Competition 2025's track for solvers that use several cores."
---

> Documentation Index
> Fetch the complete documentation index at: https://docs.cpbenchy.com/llms.txt
> Use this file to discover all available pages before exploring further.

# xcsp3-2025-parallel

## What it does

Runs your experiment the way the XCSP3 Competition 2025 runs the parallel COP track: its limits, how runs are stopped, and
its output. Results record that they followed these rules, or that you changed them.

| Wall time | CPU time | Memory | Cores | At the limit | Output |
|---|---|---|---|---|---|
| 30 min | none | 64 GiB | 4 | SIGTERM, SIGKILL 1 s later | [`XCSP3Output`](/library/xcsp3-output/) |

## Use it

```python
import cpbenchy

cpbenchy.run("instances/", solvers=["ortools", "exact"], rules="xcsp3-2025-parallel")
cpbenchy.run("instances/", solvers=["ortools"], rules="xcsp3-2025-parallel", mem_limit_mib=8192, args=["--scale", "0.05"])  # a quick try
```

```sh
cpbenchy run instances/ -s ortools -s exact --rules xcsp3-2025-parallel
cpbenchy run instances/ -s ortools --rules xcsp3-2025-parallel --scale 0.05 -m 8192   # a quick try: a twentieth of the time, 8 GiB
```

The memory limit is the competition's. On a machine with less memory, give a lower one with `-m`: the runs then record the rules as modified.

To enter the competition, build a submission with these rules: [`cpbenchy submission`](/library/submission/).

## Notes

The numbers come from the competition's own documents, cited in the file. Competitions change their
rules from year to year: check the file against the competition's call before you rely on it, and add
the next edition as a new file. [Rules](/guides/rules/) explains the format.

## Implementation

Source: https://docs.cpbenchy.com/library/xcsp3-2025-parallel/index.mdx