Coding agents find the right GPU bottleneck 70% of the time, fix it 30%

Show HN: Coding agents find the right GPU bottleneck 70% of the time, fix it 30%

Coding agents find the right GPU bottleneck 70% of the time, fix it 30% is active in the show hn launch space.

Profile

One of the authors. Some things that surprised us while running these experiments:The tasks are pulled from real merged PRs in vLLM and SGLang, so there's a known-good human solution for each one. Agents get the full codebase, the issue description, and a test harness. Pretty generous setup.What we didn't expect: the agents are genuinely good at diagnosing the problem. They read the code, find the bottleneck, describe the right fix. But then the generated code has subtle bugs. Off-by-one in kernel indexing, wrong tensor shapes, missing synchronization barriers. The kind of stuff that passes a code review at first glance but segfaults under load.The other weird result: agent rankings complete

Within the show hn launch space, the company focuses on its core offering rather than a broad product line.

It is recorded at the Show HN launch stage.

Website: ayushnangia.github.io/iso-bench-website