# GPUDrive: Data-driven, multi-agent driving simulation at 1 million FPS

*Source* · `src-kazemkhani-2024-gpudrive` · status: **n/a**

| field | value |
|---|---|
| kind | Preprint |
| authors | Saman Kazemkhani, Aarav Pandya, Daphne Cornelisse, Brennan Shacklett, Eugene Vinitsky |
| date | 2024-08-02 |
| url | https://arxiv.org/abs/2408.01584 |
| arxiv | 2408.01584 |
| accessed | 2026-10-24 |
| notes | Built on Madrona; over one million simulation steps per second; multi-agent driving agents trained on Waymo Open Motion scenarios in minutes to hours. |

## Backlinks (computed)

**cited by:** [Massively parallel GPU simulation](../technique/massively-parallel-simulation.md)

## Citation

Saman Kazemkhani, Aarav Pandya, Daphne Cornelisse, Brennan Shacklett, Eugene Vinitsky (2024-08-02). *GPUDrive: Data-driven, multi-agent driving simulation at 1 million FPS*.  https://arxiv.org/abs/2408.01584

---
Provenance: created in run `2026-10-24-techniques-sim-math`.