Skip to content

Compile take functions for uint dtypes #57632

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

Merged
merged 3 commits into from
Mar 16, 2024

Conversation

phofl
Copy link
Member

@phofl phofl commented Feb 26, 2024

Not sure why this wasn't compiled before, maybe because of wheel size? We should at least compile the exact matching, not averse to skipping the cross product with different itemsizes

@phofl phofl requested a review from WillAyd as a code owner February 26, 2024 18:18
@WillAyd
Copy link
Member

WillAyd commented Feb 26, 2024

Seems fine. Are these actually called?

@mroeschke mroeschke added the Internals Related to non-user accessible pandas implementation label Feb 27, 2024
@phofl
Copy link
Member Author

phofl commented Mar 16, 2024

Those are more or less user facing, we end up calling them if users have uint dtypes (that's how we found this in Dask)

good to merge @WillAyd?

@WillAyd
Copy link
Member

WillAyd commented Mar 16, 2024

Yep merge away

@phofl phofl merged commit 3267ba0 into pandas-dev:main Mar 16, 2024
@phofl phofl deleted the compile_other_dtypes branch March 16, 2024 17:47
@phofl
Copy link
Member Author

phofl commented Mar 16, 2024

thx

@phofl phofl added this to the 3.0 milestone Mar 16, 2024
@lithomas1
Copy link
Member

For all dtypes of the same itemsize, is it possible to e.g. view them as an int8/16/32/64, and just have one kernel that does int->int take, or is this undefined behavior?

@phofl
Copy link
Member Author

phofl commented Mar 16, 2024

Would still require a copy and this is a path that is very heavily used.

pmhatre1 pushed a commit to pmhatre1/pandas-pmhatre1 that referenced this pull request May 7, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Internals Related to non-user accessible pandas implementation
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants