Performance and Storage Improvements - #28
Open
ianfd wants to merge 31 commits into
Open
Conversation
through sprs in anndata
Update to current state of main repo
…n padding for dense arrays
…otonic fast-paths
…ntics This adds ownership-transferring methods to ArrayElem and collections to allow moving data directly into memory without caching or duplicating buffers. Also adds backend interop and nested metadata tests.
…rtial reads - Parallelize component loading for CSR/CSC matrices, DataFrames, and Collections using Rayon to maximize disk/network bandwidth. - Implement optimized read_select for sparse matrices to perform partial I/O, fetching only required data/indices for row/column slices. - Eliminate redundant casting and intermediate allocations in sparse index loading by implementing direct-to-target-type reading. - Replace sequential iter_containers with concurrent dispatch.
- added corrupt CSC/CSR tests - Cover bad indptr, length mismatch, unsorted indices, and out-of-bounds indices - some cleanup and small Zarr bugfix
Member
|
@ianfd given that you put in so much time into this PR already, would it be possible to get it green so that we can consider it, please? |
Author
|
Hey, Ofc, super time limited currently but will make these changes tonight, should make it go green. Best, |
Member
|
Not urgent! Take your time, please. |
Author
|
Updated, should work now! |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
TLDR: switched from nalgebra-sparse to sprs, allows dynamic indices of CSR/CSC
Zarr support now with direct interface to e.g. object storage
Performance statistics will be uploaded in next few days.
Performance characteristics: H5AD same/comparable in all cases tested, massive improvement observed when reading from Zarr format.