Fix worktree counting on Julia 1.13#64
Merged
Merged
Conversation
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #64 +/- ##
==========================================
- Coverage 99.20% 95.20% -4.01%
==========================================
Files 4 4
Lines 126 125 -1
==========================================
- Hits 125 119 -6
- Misses 1 6 +5 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
Contributor
Author
|
The invalidations check itself seems broken, rather than anything to do with this change. |
Contributor
Author
|
@oscardssmith you seem like you might be a good person to poke about this. |
oscardssmith
approved these changes
May 25, 2026
ChrisRackauckas
approved these changes
May 29, 2026
ChrisRackauckas
left a comment
ChrisRackauckas
left a comment
Member
There was a problem hiding this comment.
Verified: 1.13+ branch uses Base.Workqueues() which returns the current thread's IntrusiveLinkedListSynchronized{Task}, and Base.Workqueues[tidp1] continues to work via OncePerThread's getindex. CI green on nightly across linux/macos/windows including aarch64.
ChrisRackauckas
added a commit
to JuliaSIMD/LoopVectorization.jl
that referenced
this pull request
May 29, 2026
ThreadingUtilities 0.5.6 (JuliaSIMD/ThreadingUtilities.jl#64) fixes the Julia 1.13+ OncePerThread MethodError in wake_thread! that was causing every pre/nightly job to red-flag part1 and part4. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
ChrisRackauckas
added a commit
to JuliaSIMD/LoopVectorization.jl
that referenced
this pull request
May 30, 2026
* Unbreak Apple ARM tests that now pass
Several `@test_broken` / `@test_skip` gates on Apple ARM (M-series) no
longer apply with current LoopVectorization and the VectorizationBase
nested-W=1 `_vstore_unroll!` fix.
- `condstore!` masked-store tests in `ifelsemasks.jl` (lines ~626-655)
now produce matching results on Apple ARM — drop the Apple branch and
test unconditionally for both Float32 and Float64.
- `Bernoulli_logitavx`/`Bernoulli_logit_avx` with `Vector{Bool}` and an
`Int` α (`ifelsemasks.jl` line ~736) was `@test_skip`-ed but actually
passes — convert to `@test`.
- Issue #543 W=1 nested VecUnroll store test in `staticsize.jl` was
`@test_skip`-ed for v=1 on Apple ARM; with the VectorizationBase fix
it now passes for all v=1..4, n=2..8.
The remaining ARM-gated breakage in `ifelsemasks.jl` (Bernoulli with a
`BitVector` mask + Float64/Int α at lines ~715-722) and the
`tullio_issue_131` pattern in `shuffleloadstores.jl` are deeper SIMD
issues left as `@test_broken` with TODOs.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* Unbreak BitVector Bernoulli_logit tests on Apple ARM
With the companion VectorizationBase fix for dynamic-index BitArray
loads with sub-byte alignment, `Bernoulli_logitavx` and
`Bernoulli_logit_avx` now produce correct results for both
`BitVector` and `Vector{Bool}` masks on Apple M-series. The
Apple-aarch64 `@test_skip` / `@test_broken` branches are dropped.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* Fix unroll-cleanup tail bound for strided loads (tullio_issue_131)
`pointermax_index` builds the limit pointer that the unroll-cleanup
termination check is compared against. The `sub > 0` branch already
applies `incr` (when not statically known) and `stride` (when ≠ 1) to
scale the loop length into a byte/element offset, but the `sub == 0`
branch was pushing the raw `stophint` / `stopsym` straight through. For
any strided load on the unrolled axis (e.g. `arr[2i, ...]`) the cleanup
bound came out `stride×` too small, so the final tail iteration was
skipped whenever `looplen mod (UF*W) != 0`.
On Apple ARM with W=2 for Float64, this dropped the last `out_i`
iteration for every odd `out_i ≥ 3` in the tullio_issue_131 shape grid,
and analogously for Float32 with W=4. The cleanup never ran for the
1–3 trailing elements, leaving them at whatever the output array was
initialized to. Confirmed correct after fix for all
`(M, N) ∈ 4:24 × 2:8` on the tullio reproducer; `test/shuffleloadstores.jl`
goes from 4255 pass / 686 broken to 4941 pass / 0 broken on Apple M-series.
Drop the matching `@test_broken` gate and the `tullio_issue_131` comment
in `test/shuffleloadstores.jl`.
Fixes #570.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* Loosen condstore == to ≈; re-gate VB-dependent tests until VB#127 release
Two CI regressions on the previous commits:
1. `condstore!` tests in `ifelsemasks.jl` (lines 626-637) use `==` to
compare a SIMD-masked-store result against the scalar reference. On
Apple ARM the two paths can differ by a 1-ULP rounding even though
`@show`-printed values look identical (the original gate predates
that observation). Switch to `≈` — the test still catches anything
meaningful, just not artifacts of operation reordering.
2. The BitVector `Bernoulli_logit{,_}avx` tests in `ifelsemasks.jl`, the
`Vector{Bool}` + Int α variants in the same block, and the W=1
nested-VecUnroll Issue #543 testset in `staticsize.jl` all depend on
the JuliaSIMD/VectorizationBase.jl#127 fixes being available at
runtime. That PR isn't tagged yet, so CI's stock VectorizationBase
doesn't have it and the tests fail. Restore the
`Sys.ARCH === :aarch64 && Sys.isapple()` gate (as `@test_broken` /
`@test_skip`) with a comment pointing at VB#127. Once that release
lands and LV's compat is bumped, the branches can be dropped.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* Use @test_skip for BitVector Bernoulli gates (Julia-version dependent)
`@test_broken` errors on "Unexpected Pass", which makes the BitVector
+ Int α Bernoulli test fail in Julia LTS macOS aarch64 CI even though
the test happens to give the correct result there. The underlying bug
(VectorizationBase BitVector load misalignment, fixed in VB#127) is
present in some configurations but not others — Julia 1.10's older
LLVM appears to dodge it for the test inputs in question.
Switch to `@test_skip` so the gate is loose either way: when the
underlying bug bites, the test is skipped; when it doesn't, no error.
After VB#127 is released and LV's compat is bumped, the entire branch
can be dropped.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* Skip W=1 issue #543 test on all platforms (not just Apple aarch64)
The nested W=1 VecUnroll store path is picked by LoopVectorization on
different (arch, julia version) combinations than originally assumed —
the Julia nightly x86_64 macOS CI also hit it, not just Apple aarch64.
The fix is in JuliaSIMD/VectorizationBase.jl#127 and not yet in a
tagged release, so skip the v == 1 sub-case on every platform until
LV's VectorizationBase compat is bumped.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* Rerun CI on top of bumped downstream releases
* Rerun CI on top of SLEEFPirates v0.6.46+
* Bump VectorizationBase compat to 0.21.74; drop @test_skip gates
VectorizationBase v0.21.74 ships the two fixes JuliaSIMD/VectorizationBase.jl#127 added:
- `_vstore_unroll!` for the nested W=1 (scalar lane) VecUnroll path,
which `staticsize.jl`'s Issue #543 testset exercises with `v == 1`.
- The dynamic-index BitArray load misalignment fix that
`ifelsemasks.jl`'s `Bernoulli_logitavx`/`Bernoulli_logit_avx` with
`BitVector` masks depends on.
Bump LV's lower bound to `"0.21.74"` and drop the
`@test_skip ... else @test ... end` branches I added while VB#127 was
still in flight:
- `test/ifelsemasks.jl`: Bernoulli BitVector + Int α (4 tests),
Vector{Bool} + Int α (2 tests), BitVector + Float64 α (2 tests).
- `test/staticsize.jl`: the `v == 1` Issue #543 sub-case (7 entries).
Local sweep on Apple M-series with the dev'd v0.21.74:
- `test/ifelsemasks.jl`: 435/435 pass (was 430/5 broken).
- `test/staticsize.jl` Issue #543 testset: 84/84 pass (was 70/77).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* Retrigger CI to pick up ThreadingUtilities 0.5.6
ThreadingUtilities 0.5.6 (JuliaSIMD/ThreadingUtilities.jl#64)
fixes the Julia 1.13+ OncePerThread MethodError in wake_thread! that was
causing every pre/nightly job to red-flag part1 and part4.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* Remove Invalidations CI workflow
The SnoopCompileCore-based invalidations check has been broken since the
SCPrettyTablesExt FieldError upstream regression and has been red across
all recent PRs. The signal it produced (regressions in method-table
invalidation count) hasn't been actionable for this repo; removing the
workflow rather than keeping a perma-red check.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
1 task
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
3 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.
Rather trivial. Fixes #63.
Can confirm that all tests, and Polyester.jl tests, now pass on 1.13-rc1.