Add regression test for GPU pred=3 multi-sample TIFFs (#1479)#1481
Merged
Conversation
The GPU _fp_predictor_decode_kernel handles multi-sample float TIFFs correctly, but predictor=3 multi-sample was not exercised by CI. Only predictor=2 multi-sample had GPU coverage. Add parametrized round-trip tests over samples in (3, 4) and dtype in (float32, float64) to lock in current behaviour.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Closes #1479.
Summary
_fp_predictor_decode_kerneldecodes multi-sample float TIFFs (predictor=3, samples=3 or 4, float32 or float64) correctly. Manual round-trip testing confirms it.test_predictor_multisample.pycovered GPU predictor=2 multi-sample but not predictor=3. CI was not exercising the path.samples in (3, 4)anddtype in (float32, float64)that write viato_geotiff(..., predictor=3, tiled=True, tile_size=32)and assert GPU decode is bit-exactly equal to CPU decode and to the source.This is a coverage gap, not a bug fix. No source changes.
Test plan
pytest xrspatial/geotiff/tests/test_predictor_multisample.pypasses (14 tests, 4 new)gpu_onlyskipif so CPU-only CI is unaffected