perf(parser): make Modifiers::accessibility branchless

[overlookmotel]

Modifiers::accessibility is called in many places. Previously it was 10 instructions, including 2 unpredictable branches.

Reduce it to 4 instructions and no branches, by using an 8-byte lookup table.

The code is verbose, but almost all of it is compile-time constants, which boil down to nothing but the 8-byte table. The code could be made shorter, but only by resorting to unsafe code. The version in this PR contains no unsafe code, and produces the same optimal assembly as an unsafe version.

Before:

accessibility:
        mov     al, 2
        test    dil, 8
        jne     .LBB1_3
        mov     al, 1
        test    dil, 4
        jne     .LBB1_3
        xor     eax, eax
        test    dil, 2
        sete    al
        lea     eax, [rax + 2*rax]
.LBB1_3:
        ret

After:

accessibility:
        shr     edi
        and     edi, 7
        lea     rax, [rip + LUT::h8cdda7c57c71092a]
        movzx   eax, byte ptr [rdi + rax]
        ret

LUT::h8cdda7c57c71092a:
        .ascii  "\003\000\001\001\002\002\002\002"

+0.1% - +0.2% improvement on parser benchmarks. But CodSpeed doesn't take into account the cost of branch misprediction which we now avoid, so it's likely an underestimate.

[overlookmotel]

• perf(parser): move kinds field of Modifiers to first #20833
• perf(parser): add Modifiers::contains_accessibility #20832
• perf(parser): make Modifiers::accessibility branchless #20827 👈 (View in Graphite)
• perf(parser): add #[inline] to trivial modifier methods #20826
• perf(parser): defer calculation of Modifier spans #20823 : 1 other dependent PR (#20824 )
• perf(parser): remove branch from parse_modifiers #20821
• perf(parser): reduce Kind to ModifierKind conversions in eat_modifiers_before_declaration #20820
• refactor(parser): shorten code handling illegal modifiers #20781
• perf(parser): fast path for no syntax errors when checking modifiers #20748
• perf(parser): remove a lookahead from eat_modifiers_before_declaration #20755
• refactor(parser): nest modifiers types in modules #20754
• refactor(parser): re-order modifiers code #20753
• perf(parser): store Modifiers on stack #20742
• perf(parser): add Modifiers::get method #20741
• refactor(parser): rename ModifierKinds::empty to none #20740
• refactor(parser): rename ModifierFlags to ModifierKinds #20739
• refactor(parser): simplify ModifierFlags #20738
• refactor(parser): import from std at top level #20737
• refactor(parser): remove Modifier::is_static() #20736
• refactor(parser): remove dead code from modifier parsing #20735
• main

---

How to use the Graphite Merge Queue

Add either label to this PR to merge it via the merge queue:

• 0-merge - adds this PR to the back of the merge queue
• hotfix - for urgent changes, fast-track this PR to the front of the merge queue

You must have a Graphite account in order to use the merge queue. Sign up using this link.

An organization admin has enabled the Graphite Merge Queue in this repository.

Please do not merge from GitHub as this will restart CI on PRs being processed by the merge queue.

This stack of pull requests is managed by Graphite. Learn more about stacking.

[Copilot]

Pull request overview

This PR optimizes TypeScript accessibility-modifier detection in the parser by making Modifiers::accessibility branchless and delegating the work to a new ModifierKinds::accessibility implementation that uses a packed lookup table.

Changes:

• Replaced branchy Modifiers::accessibility checks with a call to ModifierKinds::accessibility.
• Added a new branchless ModifierKinds::accessibility using a packed LUT (u64) for fast decoding.
• Added a small const fn min helper for const-eval computations used by the LUT setup.

[codspeed-hq]

Merging this PR will degrade performance by 4.46%

❌ 1 regressed benchmark
✅ 47 untouched benchmarks
⏩ 8 skipped benchmarks¹

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Simulation	mangler[RadixUIAdoptionSection.jsx_keep_names]	19.4 µs	20.4 µs	-4.46%

---

_{Comparing om/03-28-perf_parser_make_modifiers_accessibility_branchless (5995339) with om/03-28-perf_parser_add_inline_to_trivial_modifier_methods (2208114)}

Footnotes

1. 8 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[Copilot]

Pull request overview

Copilot reviewed 1 out of 1 changed files in this pull request and generated 1 comment.

[camc314]

Merge activity

• Mar 29, 3:33 PM UTC: The merge label '0-merge' was detected. This PR will be added to the Graphite merge queue once it meets the requirements.
• Mar 29, 3:40 PM UTC: The Graphite merge queue removed this pull request due to removal of a downstack PR #20820.
• Mar 29, 3:40 PM UTC: camc314 added this pull request to the Graphite merge queue.
• Mar 29, 3:41 PM UTC: The merge label '0-merge' was removed. This PR will no longer be merged by the Graphite merge queue
• Mar 29, 3:42 PM UTC: camc314 added this pull request to the Graphite merge queue.
• Mar 29, 4:19 PM UTC: The Graphite merge queue removed this pull request due to downstack failures on PR #20820.
• Mar 29, 4:51 PM UTC: The merge label '0-merge' was removed. This PR will no longer be merged by the Graphite merge queue
• Mar 29, 4:51 PM UTC: graphite-app added this pull request to the Graphite merge queue.
• Mar 29, 4:52 PM UTC: Merged by the Graphite merge queue.