News for hackers. Stuff that matters.

A comment on Show HN: Flash Attention in ~100 lines of CUDA. Parent comment

Re:Show HN: Flash Attention in ~100 lines of CUDA by queuebert on Saturday March 16, 2024 @08:35PM (#39730758)
As a person who finds CUDA extremely easy to write and integrate, what does Triton have to offer?
[ Parent | on HN ]
Re:Show HN: Flash Attention in ~100 lines of CUDA by whimsicalism on Saturday March 16, 2024 @10:09PM (#39731325)
block level rather than thread level programming, automatic optimization across hyperparameters, makes it much easier to write fast kernels
[ Parent | on HN ]