Allow blocks smaller than full shape along reprojected dimensions and fix calculation of co-addition footprint in broadcasted parallelization#613
Open
astrofrog wants to merge 11 commits into
Conversation
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #613 +/- ##
==========================================
+ Coverage 90.05% 90.18% +0.12%
==========================================
Files 51 51
Lines 2293 2312 +19
==========================================
+ Hits 2065 2085 +20
+ Misses 228 227 -1 ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
astrofrog
changed the title
astrofrog
force-pushed
the
blockwise-reproject
branch
from
5d5330d to
e3ccdb8
Compare
astrofrog
changed the title
astrofrog
changed the title
Member
Author
|
pre-commit.ci autofix |
astrofrog
force-pushed
the
blockwise-reproject
branch
from
6381d0c to
9293891
Compare
…mensions for non_reprojected_dims reprojection, tiling each plane so the coordinate transform memory stays bounded, and route dask array inputs through map_blocks so each block is reprojected from the input directly
…ingle_block function inside the dispatcher instead of two module-level helpers so the diff against main stays small and easy to review
…aps, since views keep the parent's unadjusted offset and were silently reprojected from the wrong file region, and pass views by reference instead
…ith one chunk per non-reprojected slice, each routed through a delayed task so blockwise fusion cannot recompute it per output tile, restoring per-slice streaming for dask inputs instead of materializing them to a temporary memmap and letting distributed schedulers ship each task only its own slice
…' and it is chunked below one slice along the reprojected dimensions, so streaming reprojection cores only compute the input chunks each output tile touches and never need to hold a whole slice, while slice-chunked inputs are still materialized exactly once per slice
…sions that are neither 1 nor the full extent instead of silently reinterpreting them as one slice per block
…_blocks call so edge blocks are reprojected at their true size instead of being computed at the full block size and truncated, which also avoids an irregular trailing chunk in dask results
…-in for a custom lazy array, since the dispatcher no longer needs it
…d document that it prevents dask from hashing the whole buffer to name the array, which for a memmap silently loads the entire file into memory
…sions in the sub-tiling comments, docstrings and error message, since the reprojected dimensions are not necessarily two-dimensional or celestial
astrofrog
force-pushed
the
blockwise-reproject
branch
from
dcc4fac to
67f27d7
Compare
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
1 participant
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.
WIP