Refactor num_flops and fix for loopy kernels #668
Closed
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…ernel does that, but we count the FLOPS of the local kernel here. This changes the result of the extrusion kernel but not its FLOP count I think.
sv2518
commented
Jul 11, 2022
Contributor
Author
|
Currently the counting is spewing a lot of warnings in loopy and silencing doesn't work as I would expect (possibly my fault). I asked the Loo.py people here inducer/loopy#645. |
connorjward
requested changes
Jul 11, 2022
connorjward
approved these changes
Jul 11, 2022
Contributor
Author
|
Thanks, Connor! I will waiting with merging this until the warnings are shut up properly. |
Collaborator
|
Closing as we are moving PyOP2 into Firedrake. There is a Firedrake PR that references this. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This a fix for the flop counting. If we turn flop counting on in Firedrake without this fix, all tests are failing see actions run on this commit firedrakeproject/firedrake@fae62c4. The commit later pulls this PyOP2 branch in.
I also refactored the code a tiny bit and made
num_flopsa method of the specialized local kernels rather than the generic to avoid some of theisinstancespecial casing we had before.The coffee flop counting is fixed now too.
There is something funny going on with layer args in the flop counting algorithm that I don't understand yet (see failing tests @ https://github.com/firedrakeproject/firedrake/runs/7272623399?check_suite_focus=true. (EDIT: I had to fix the layer arg in the loopy kernels because we usually pass a value for that in from the global kernel)
Order of merging: firedrakeproject/tsfc#279, then this one, then firedrakeproject/firedrake#2490