Docs: make ThreadFusion's opt-in semantics explicit (matches FKL default, per Oscar's guidance) - #3
Merged
morousg merged 2 commits intoJun 11, 2026
Conversation
thread_fusion was ALWAYS opt-in (default False, matching FKL's own TransformDPP<> = TF::DISABLED default), but the docstring/README sold it too enthusiastically. Now both state explicitly: disabled by default, matching upstream, because per Oscar it only improves performance in a small set of cases — benchmark your pipeline before enabling.
C++-only probe using fk::CircularTensor exactly as designed (float, COLOR_PLANES=3, NewestFirst, TensorSplit write). Values come out CORRECT (rotation 5/4/3) but compute-sanitizer reports 33 invalid reads/writes per run on unfixed headers: update's 1-plane read + copy's BATCH-plane read sum to BATCH+1 launched planes. Sanitizer-clean with FKL PR #252 applied. This is the evidence posted on FusedKernelLibrary#250.
morousg
approved these changes
Jun 11, 2026
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Documentation-only follow-up after Oscar's ThreadFusion feedback.
thread_fusionwas always opt-in (defaultFalse, matching FKL's ownTransformDPP<>=TF::DISABLEDdefault), but the docstring and README oversold it ("best for wide images..."). Both now state the upstream guidance explicitly:No behavior changes; no example or test enables it implicitly (the only
thread_fusion=Truein the repo is the dedicated correctness suite that verifies TF output is identical to scalar).