Nanops numba implementation #62536
Open
Nanops numba implementation #62536
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
eoincondron
force-pushed
the
nanops-numba-implementation
branch
3 times, most recently
from
66f52c7 to
babae54
Compare
eoincondron
force-pushed
the
nanops-numba-implementation
branch
from
babae54 to
03168ed
Compare
…g parallel chunking for large arrays
…s_numba - Tests all 9 private methods prefixed with underscore - 37 test cases organized in 8 test classes - Comprehensive coverage of Numba-accelerated reduction operations - Tests edge cases: NaN handling, empty arrays, masks, different dtypes - Uses pytest fixtures and parameterization to avoid code duplication - Tests NumbaList usage for parallel processing - Uses pandas._testing for consistent assertion helpers - All tests pass in pandas-dev environment Functions tested: - _get_initial_value: Finding first valid values in arrays - _nb_reduce_single_arr: Single array reduction operations - _nullify_below_mincount: Minimum count validation - _reduce_empty_array: Empty array handling - _chunk_arr_into_arr_list: Array chunking for parallel processing - _nb_reduce_arr_list_in_parallel: Parallel reduction operations - _reduce_chunked_results: Combining chunked results - _cast_to_timelike: DateTime/timedelta type casting - _nanvar_std_sem: Variance/standard deviation/standard error calculations 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <noreply@anthropic.com>
eoincondron
force-pushed
the
nanops-numba-implementation
branch
from
03168ed to
9c7f945
Compare
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Outdated suggestions cannot be applied.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.
Ref #62449
The performance of array reductions in
nanops/bottleneckcan be significantly improved upon for large data using numba. The improvements are due to two factors:Although the added code is fairly complex, it provided a central, unified piece of code built from scratch covering the different reductions across different data types, array classes, skipna toggle, masked arrays etc. , potentially replacing code existing across multiple modules and, in the case of
bottleneck, code that lives in a different repository.It currently covers
nan(sum|mean|min|max|var|std|sem)and should be easily extensible. I am seeking code review before bottoming it out completely, so as not to waste effort.This screenshot demonstrates a potential 4x improvement on a DataFrame of 10-million rows and 5 columns of various types.
I am running the code on a features branch, and all unit tests for the feature branch are passing locally.
https://github.com/eoincondron/pandas/tree/nanops-numba-implementation
The hardware is a new MacBook Pro with 8 cores.
The performance is still slightly better at 1-million rows and is even greater at larger magnitudes (8x at 100 million rows).
The caveat is that all JIT-compilation is already completed.
I have carried out a more comprehensive performance comparison and these results hold up.
Similarly to
bottleneck, these codepaths can be toggled on and off.