[connortsui20]

Summary

Remove third-party timing baselines, old decode-first comparison paths, and one exact duplicate. Keep benchmarks for distinct Vortex operations and representations.

Changes

• Reduce the integer Widen target from 46 cases to 17. Keep the operations on Widen arrays and measure decoding separately, instead of repeating every operation on canonical and explicitly decoded inputs.
• Remove the expanded-index take baselines, decode-then-compare baselines, and the benchmark that manually reifies constant run lengths. Keep the maintained entry points and their distinct input representations.
• Remove patterns_i128(Random), which duplicates random_i128(0.5) with the same values and mask.
• Remove Arrow and std timing baselines and their unused dependencies. Keep the row_encode --output-sizes comparison and the documented local allocation baselines.
• Update the affected sharding estimates in scripts/codspeed_bench_counts.json, preserving the new decimal Widen target. These estimates are not a fresh inventory of every simulation benchmark.

The restored integer arithmetic, Widen slice/chunked, bitpacked Lt, and mask iteration cases remain. The exclusions from #11176 also remain in effect.

Validation

The original revision reported targeted Clippy and simulation builds, shard-planner tests, spelling checks, and pinned-nightly formatting. Those checks have not been rerun after rebasing onto current develop. No local checks were run during this cleanup. The pre-commit and pre-push hooks were bypassed as requested. CI validation is pending.

[connortsui20]

I brought back a few

[codspeed]

Merging this PR will degrade performance by 4.24%

1 warning

⚠️ Walltime results may be unreliable

Some benchmarks used the Walltime instrument outside CodSpeed Macro Runners, which can make the results inconsistent.

Learn more about Macro Runners

See affected benchmarks

⚡ 3 improved benchmarks
❌ 5 regressed benchmarks
✅ 1740 untouched benchmarks
🆕 17 new benchmarks
⏩ 611 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	CPU Simulation	bitwise_and_vortex_buffer[128]	4.1 µs	5.8 µs	-29.56%
❌	CPU Simulation	patched_take_10_stratified	55.1 µs	69.4 µs	-20.68%
❌	CPU Simulation	random_i64[0.95]	129.1 µs	155.5 µs	-16.99%
❌	CPU Simulation	primitive_i64_vortex	72.3 µs	86.1 µs	-16.01%
❌	CPU Simulation	fragmented_runs[7]	111.1 µs	126.5 µs	-12.15%
⚡	CPU Simulation	patterns_i128[Runs]	99.2 µs	74.3 µs	+33.59%
⚡	CPU Simulation	random_i128[0.5]	123.1 µs	95.4 µs	+29.1%
⚡	CPU Simulation	random_i128[0.8]	149.1 µs	124.5 µs	+19.82%
🆕	CPU Simulation	add_canonical	N/A	929.3 µs	N/A
🆕	CPU Simulation	add_constant	N/A	533.8 µs	N/A
🆕	CPU Simulation	add_widen	N/A	598.8 µs	N/A
🆕	CPU Simulation	between_in_range	N/A	136.4 µs	N/A
🆕	CPU Simulation	cast_f64	N/A	349.6 µs	N/A
🆕	CPU Simulation	chunked	N/A	1.5 ms	N/A
🆕	CPU Simulation	compare_in_range	N/A	161.5 µs	N/A
🆕	CPU Simulation	decode	N/A	342.8 µs	N/A
🆕	CPU Simulation	fill_null_out_of_range	N/A	1.2 ms	N/A
🆕	CPU Simulation	filter	N/A	325 µs	N/A
🆕	CPU Simulation	interleave	N/A	1.3 ms	N/A
🆕	CPU Simulation	min_max	N/A	130.3 µs	N/A
...	...	...	...	...	...

ℹ️ Only the first 20 benchmarks are displayed. Go to the app to view all benchmarks.

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing ct/stoic-rubin-t4f27x (3cd9a97) with develop (24cd0ea)}

Footnotes

1. 611 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. ↩