Skip to content

perf(codegen/transform): #6812 — multi-group clone (w9) + inliner PutValueSet arm (w6), both beat node #6868

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
proggeramlug merged 5 commits into from
Jul 27, 2026
+378 −202
Merged

perf(codegen/transform): #6812 — multi-group clone (w9) + inliner PutValueSet arm (w6), both beat node #6868

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
f573f1c
perf(codegen): #6812 w9 — multi-group whole-loop write clone
Jul 26, 2026
c5cd16e
docs: #6812 w9 multigroup changelog fragment + matrix row
Jul 26, 2026
4044184
fix(transform): #6812 w6 — inliner descends into PutValueSet references
Jul 26, 2026
1a9c6fc
docs: #6812 w6 row + fragment addendum
Jul 26, 2026
68f38e7
docs: PR-keyed fragment name; summary count + w6 baseline aligned (Co…
Jul 27, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions changelog.d/6868-w9-multigroup.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1 @@
perf(codegen): #6812 — the whole-loop write clone matches multi-group bodies: up to two (alias, temps, writes) groups over distinct arrays, the shape the inliner produces for a write helper applied to parallel arrays. Each group is monomorphic on its own array (one preflight guard call each, AND-gated fast nest, interleaved stores, spill lanes included); dynamic bounds stay single-group. w9 (two shapes through one helper): 27 ms → ~5-8 ms vs node ~6 ms. Also fixes the inliner's missing PutValueSet arm — calls in a write's RHS were invisible to inlining, stranding loops like `o.c = pick(r) + i` on the generic path; the hoisted call result now lands as a temp the clone matcher admits. w6 (call in body): ~100 ms → 5-7 ms, beating node (~7 ms).

@coderabbitai coderabbitai Bot Jul 27, 2026

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win

🧩 Analysis chain

🏁 Script executed:

#!/bin/bash
set -euo pipefail

echo "== files =="
git ls-files | grep -E '(^changelog\.d/6868-w9-multigroup\.md$|^docs/object-write-matrix\.md$)' || true

echo
echo "== changelog excerpt =="
if [ -f changelog.d/6868-w9-multigroup.md ]; then
  cat -n changelog.d/6868-w9-multigroup.md
fi

echo
echo "== docs table excerpt =="
if [ -f docs/object-write-matrix.md ]; then
  sed -n '1,75p' docs/object-write-matrix.md | cat -n
fi

echo
echo "== related benchmark mentions =="
rg -n "w6|w9|27 →|~5-8|~0\.9|BEATS node|Ties/beats|at or better|14 of 18" -S .

Repository: PerryTS/perry

Length of output: 36419

Align benchmark verdicts with the reported ranges. The reported w6 Node baseline of ~7 ms overlaps Perry’s 5–7 ms range, while w9 is only documented as “ties/beats” with a pending idle run. Use conservative wording for w6 until a strict median is established, and update the aggregate label to include non-strict wins.

  • changelog.d/6868-w9-multigroup.md#L1-L1: replace w6 “beating node” with conservative wording or provide the exact supporting median.
  • docs/object-write-matrix.md#L43-L46: change the w6 verdict from unconditional “BEATS node” to “TIES/BEATS” unless exact measurements justify the stronger claim.
  • docs/object-write-matrix.md#L58-L63: rename the aggregate to count wins or ties, e.g. “At node performance or better,” since w9 is still tied/beating with exact ratio pending.
📍 Affects 2 files
  • changelog.d/6868-w9-multigroup.md#L1-L1 (this comment)
  • docs/object-write-matrix.md#L43-L46
  • docs/object-write-matrix.md#L58-L63
🤖 Prompt for AI Agents 
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@changelog.d/6868-w9-multigroup.md` at line 1, Align benchmark wording with
the available evidence: in changelog.d/6868-w9-multigroup.md:1, replace the w6
“beating node” claim with conservative wording or an exact supporting median; in
docs/object-write-matrix.md:43-46, change the w6 verdict to “TIES/BEATS” unless
measurements justify “BEATS”; and in docs/object-write-matrix.md:58-63, rename
the aggregate label to reflect wins or ties, such as “At node performance or
better.”

Loading
Loading