Trending storyWatching
NVIDIA Warp and MjWarp let you run 2,048 robot simulations in parallel on GPU
1 report1 sourceupdated 2 days ago
What happened
Summary
NVIDIA 发布了 MjWarp，一个基于 Warp 的 GPU 加速版 MuJoCo。经典 MuJoCo 跑在 CPU 上，靠多核并行；MjWarp 跑在 GPU 上，能同时仿真最多 2048 个世界。这对强化学习这类需要大量采样的任务很关键——数据直接留在 GPU 上，不用来回搬。文章用 SO-101 机械臂演示了迁移流程，但没给具体加速倍数，所...
Coverage
Follow the reports to see the story from different sides.
Sep 24
- Hugging Face BlogNVIDIA Warp and MjWarp let you run 2,048 robot simulations in parallel on GPU
NVIDIA released MjWarp, a GPU-accelerated version of MuJoCo built on Warp. Classic MuJoCo runs on CPU and parallelizes across cores; MjWarp runs on GPU and can simulate up to 2,048 worlds at once. This matters for learning workloads like RL that need massive sampling—data stays on the GPU. The post walks through migrating an SO-101 arm, but doesn't give exact speedup numbers.
Heat over time
Not enough continuous observations to draw a trend yet.