Research map / Multi-agent systems and games / Cooperative multi-agent RL
WarpDrive: Fast End-to-End Deep Multi-Agent Reinforcement Learning on a GPU
Tian Lan, Sunil Srinivasa, Huan Wang, Stephan Zheng
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| scale number of agents from 5 to 1000 in discrete Tag | WarpDrive · Discrete Tag (2000 envs) | training iterations per second: 1.3 million -> 0.58 million | Hurt |
| run RL on a single A100 GPU instead of a 16-CPU node | WarpDrive · COVID-19 economic simulation | steps per second: 24x more steps per second | Improved |
| run continuous Tag training on GPU rather than 16-CPU node | WarpDrive · Continuous Tag, 12 agents, 60 envs | training speed: 5x faster | Improved |
| use the WarpDrive CUDA action sampler instead of the PyTorch operator | WarpDrive · Discrete Tag, 2000 envs | samples per second: 5 million -> 18 million; 3.6x faster | Improved |
| simulate environments on GPU instead of single-CPU NumPy | WarpDrive · Discrete Tag per environment | throughput: at least 50x faster | Improved |
Closest papers by meaning
- JaxMARL: Multi-Agent RL Environments and Algorithms in JAX (NeurIPS 2024 · similarity 0.88)
- NAVIX: Scaling MiniGrid Environments with JAX (NeurIPS 2025 · similarity 0.87)
- Kinetix: Investigating the Training of General Agents through Open-Ended Physics-Based Control Tasks (ICLR 2025 · similarity 0.87)
- GPUDrive: Data-driven, multi-agent driving simulation at 1 million FPS (ICLR 2025 · similarity 0.87)
- Dr. MAS: Stable Reinforcement Learning for Multi-Agent LLM Systems (NeurIPS 2026 · similarity 0.87)
- Staggered Environment Resets Improve Massively Parallel On-Policy Reinforcement Learning (NeurIPS 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
