Merged
Conversation
Contributor
Author
|
I think the array API test suite is starting to share my viewpoint on the compiler ("compilation step") performance:
|
tylerjereddy
added a commit
to tylerjereddy/pykokkos
that referenced
this pull request
Sep 29, 2022
* as discussed briefly on Slack, this is a draft branch for precompiling a subset of the ufuncs to avoid JIT slowdown errors like the one described here for the `hypothesis`-driven array API conformance test suite: kokkos#97 (comment) * for now, this focuses on `pk.double` type in a single dimension, simply because all ufuncs should support that for pre-compilation at this time (we can broaden the dim/type pre-compile as ufuncs gain more flexibile type/dim support) * one major issue I see here is that I don't think the pre-compilation will actually do anything useful yet because of how the `pk_cpp` directory structures work on a per-module non-shared hierarchy basis:
NaderAlAwar
approved these changes
Oct 1, 2022
tylerjereddy
added a commit
to tylerjereddy/pykokkos
that referenced
this pull request
Oct 4, 2022
* this is in part a copy of the NumPy patch applied at: numpy/numpy#22365 and is related to this failure we were seeing with `hypothesis`: kokkos#65 (comment) * the standard section related to this is: https://data-apis.org/array-api/latest/future_API_evolution.html#versioning * unpin `hypothesis` in our CI as a result; to be fair, this is the solution @ShahzadUmair wanted from the start I think * the updated testing behavior required me to cherry-pick some data type info improvements from kokkosgh-97
* changes needed to pass `test_isfinite()` in the array API conformance suite, and turn this test on in CI * this required me to write substantially more workunits/kernels for that ufunc, and to fill in some more gaps in the type info system, including the scalar `int` conversion for single-element views
b59a9fd to
c831d38
Compare
Contributor
Author
|
I rebased/updated/force-pushed. This already has an approval, but I'd like to see CI flush through again after the update (maybe be some delay with actions at the moment because of a degraded performance report: https://www.githubstatus.com/). |
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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.
changes needed to pass
test_isfinite()in the array API conformance suite, and turn this test on in CIthis required me to write substantially more workunits/kernels for that ufunc, and to fill
in some more gaps in the type info system, including the scalar
intconversion for single-element views