-
Notifications
You must be signed in to change notification settings - Fork 23
perf: run ablations for the paper #1408
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.
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
base: main
Are you sure you want to change the base?
Conversation
1c0438a
to
f06786c
Compare
14817ba
to
f2c5e75
Compare
FNO case is heavily dominated by FFTs so nothing much we can do there, but the gradients with passes enabled being consistently slower seems off |
I assume this is all straight line (no control flow like while), and we have no fft specific optimizations yet in which case this also becomes a question of how well xla’s allocation and other passes may recognize certain patterns |
But also concurrently we built the passes for the primal and post jax differentiation , so we may be missing optimizations relevant to enzyme generated programs |
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
139af10
to
70f90b7
Compare
No description provided.