[Autoloop: tsb-perf-evolve]

[github-actions]

tsb-perf-evolve — Performance Optimization: sortValues

🤖 This PR is maintained by Autoloop. Each accepted iteration adds a commit to this branch.

Program Goal: Minimize tsb_mean_ms / pandas_mean_ms ratio for Series.sortValues on n=100k numeric data. Lower is better; < 1.0 means tsb beats pandas.

Current Best Metric: 21.048 (tsb=112.50ms / pandas=5.34ms)

Issue: #189

Iteration 35 (latest)

Three micro-optimizations on the 8-pass LSD radix sort:

1. Merged histogram: All 8 histogram passes now accumulate inline during the partition/init loop — eliminates a separate O(n) scan over ~300k Uint32 values per call.
2. Stride counter: si += 3 replaces i * 3 in scatter and gather loops — avoids 800k multiplies per call.
3. RangeIndex fast path: When the index is a default 0-based RangeIndex, bypass 100k bounds-checked at() calls from index.take(perm) by constructing the output Index directly.

Branch: autoloop/tsb-perf-evolve

Warning

Firewall blocked 1 domain

The following domain was blocked by the firewall during workflow execution:

• releaseassets.githubusercontent.com

To allow these domains, add them to the network.allowed list in your workflow frontmatter:

network:
  allowed:
    - defaults
    - "releaseassets.githubusercontent.com"

See Network Configuration for more information.

Generated by Autoloop · ● 5M · ◷