As a person who finds CUDA extremely easy to write and integrate, what does Trit... | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

		queuebert on March 17, 2024 \| parent \| context \| favorite \| on: Show HN: Flash Attention in ~100 lines of CUDA As a person who finds CUDA extremely easy to write and integrate, what does Triton have to offer?

whimsicalism on March 17, 2024 [–]

block level rather than thread level programming, automatic optimization across hyperparameters, makes it much easier to write fast kernels

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact