Repository navigation
Remove normalization buffers that amplify multicore image preparation cost - #1107
Merged
Merged
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Per-plane image normalization creates cast planes and divided planes before stacking them into another complete buffer. Under four workers, WoundHealing image-request preparation takes a median 226.54 ms per assignment, versus 106.93 ms for image processing. The exact reconstructed invocation proves the loaded source is reused; no source admission guards are loosened.
Delegate plane normalization to the existing ArrayBridge memory operation owner, and keep divisor validation and scale proofs in one existing OpenHCS intensity authority. Require ArrayBridge 0.3.8 and pin its merged implementation (ArrayBridge PR #9). Mutable arrays allocate one owned output buffer; immutable backends retain the original recipe. Producer pixels remain unchanged.
The bounded four-fork saved-source replay reduced the three-assignment critical path from 1.174 s to 0.377 s with exact pixels and metadata. This is diagnostic boundary evidence, not a full-pipeline speedup claim. Validation includes 252 original-owner comparisons, genuine WoundHealing RGB inputs, three CuPy controls, the existing OpenHCS normalization checks, and the full ArrayBridge suite (242 passed, 17 skipped); external lint, format, and type checks passed. Matched production receiving is complete: all 96 assignment comparisons passed. Full output-complete execution medians are 7.908462 s with one worker and 3.199810 s with four workers, giving 2.471542x actual scaling with the new one-worker denominator. The prior published four-worker result was 5.207406 s. Raw reports, all eight receipts/progress streams, diagnostic controls and hashes are archived in
benchmark/results/normalization_scaling_receiving_20261008. External startup and scientific qualification are outside these clocks; saving, exports, publication and finalization are included. This bounded result does not replace the full thirty-workflow manuscript figures or complete the minimum 3x scaling goal.Closes #1106.