5 ms·
> propose, implement, measure, keep the wins Pretty much what I did to let Codex with gpt5.4xhigh improve my fairly complex CUDA kernel which resulted in 20x t
by osti 5mo ago
> propose, implement, measure, keep the wins
Pretty much what I did to let Codex with gpt5.4xhigh improve my fairly complex CUDA kernel which resulted in 20x throughput improvement.
- hackyhacky 5mo agoConcretely, what interesting changes did it make to achieve such a significant improvement?
- osti 5mo agoA lot of it was beyond me, but this was all the branch names for all the stuff it tried, most of it unsuccessful of course. About 10x perf improvement came from architectural changes, and then 2x from micro optimizations. https://pastebin.com/eac0SAYg https://pastebin.com/eac0SAYg