ForgeStencil automates stencil optimization and deployment across 100+ real HPC apps
新智元 · wechat · 2026-08-04
ForgeStencil uses two agents to automate stencil optimization and deployment
MiniMax and OpenBMB released ForgeStencil, an open-source system that claims to be the first full “automatic research + automatic deployment” loop for stencil optimization.
The system combines KernelAgent and AppAgent. KernelAgent searches for fast kernels by generating and testing candidates across different stencil shapes, grid layouts, precision settings, and scheduling strategies. AppAgent then integrates those optimized kernels back into real applications by locating hotspots, extracting stencil loops, validating correctness with the app’s own tests, and deploying the result end to end.
According to the article, ForgeStencil optimized 100+ real industrial and scientific applications in one week with zero human expert intervention. On benchmarked stencil workloads, it reports geometric-mean speedups of 2.35× in fp32, another 1.95× on mixed-precision fp16 read/write, and 1.34× on the hardest variable-coefficient stencil cases. For real applications, it cites 3.86× on hypre Red-Black GS, 2.47× on FDTD, 5.78× on a reactor neutronics kernel, and additional gains in seismic imaging and medical imaging workloads.
Related event: OpenBMB Releases ForgeStencil for Automated Code Optimization(2 posts)→
More from coding & agent
- Dev Builds Free 'Radar' Scanner to Catch Security Gaps in AI-Generated Websites — indishere · 2026-08-04
- Include Standard Libraries in RL Environments to Stop AI Agents Reinventing the Wheel — cephaloform · 2026-08-04
- From GANs to Agent Swarms: An AI Vet Reencounters 'Adversarial Collapse' 12 Years Later — bingxu_ · 2026-08-04
- Building Social Media Agents: API Approvals Are the Real Bottleneck — arslan2012 · 2026-08-04
- Qwen 3.8 Coding Test: Nearly Matches K3 at Half the Price — bindureddy · 2026-08-04
- Overcoming State Loss in Long-Horizon Agents: New Framework Boosts Accuracy — Ziyu Ma · 2026-08-04