New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Meaning of divisible_by_16 configuration for triton kernel compilation #4834

Open

thumbe3 opened this issue Oct 1, 2024 · 0 comments

thumbe3 commented Oct 1, 2024

triton/python/triton/tools/compile.py

Line 109 in 49266aa

divisible_by_16 = [i for i, h in hints.items() if h == 16]

In this line what exactly does divisible_by_16 mean? Does it mean tensor length should be divisble by 16 elements or 16 bytes?

Both cases seem to be really hard constraints. Lets say I want to vector load a float in CUDA I can load float2 with just 8 byte alignment.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment