[connortsui20]

Summary

• Tracking Issue: Epic: Row-oriented scalar functions #9128

Replaces the primitive comparison migration in #9979, using the shared executor from #10675. This PR depends on #10675 and has been replayed onto its rebased head 7c47d4bcfee3370684c1b22de999e61f66628afa.

Reopened as a draft. The controlled measurements showed substantial AVX2 gains as well as ARM and AVX-512 regressions. They justify further work before merging, but do not establish that this migration has no benefit. #10741 isolates some float32 packing gains without this migration, while #10742 explores a smaller validity change. Neither is a complete replacement for every observed benefit here.

Changes

The private row kernel selects the primitive type and comparison operator. The framework owns borrowed inputs, constant handling, strict validity, and packed Boolean output. Removes the primitive operand wrapper and handwritten primitive kernels while preserving the public Binary identity and Vortex's floating-point total ordering.

Native Graviton3 measurements showed regressions when this migration used only #10673. Removing only the fused-execution patch while keeping all three foundations also slowed the tested comparisons. The dependency on #10675 keeps that execution work available when this migration merges.

Existing comparison tests exercise the shared executor. Source range-diff review confirms that the migration patch is unchanged by the rebase. Tests, formatting, and benchmarks were not rerun on the rebased head, so earlier measurements are not validation of this revision.

[codspeed]

Merging this PR will regress 17 benchmarks

4 warnings, 1 note

⚠️ 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

⚠️ Unable to measure the execution of 2 benchmarks

These benchmarks measured no execution time, so their results are not comparable and count as unchanged. This usually happens when the compiler removes the code under test.

Prevent compiler optimizations

See affected benchmarks

⚠️ 18 benchmarks are dominated by system calls

System calls cannot be simulated consistently and are left out of the measure. The reported value understates the real cost. Switch to the Walltime instrument to measure them.

Learn more about system calls

See affected benchmarks

⚠️ Base and head ran on different runner types

Some benchmarks with significant performance changes were compared across different runner types, so the changes might not come from the code. Run base and head on the same runner type, or use CodSpeed Macro Runners for more stable results.

Learn more about Macro Runners

Open the report in CodSpeed to investigate

ℹ️ 40 benchmarks have no flamegraph

The flamegraph wasn't generated: the symbols of the profiling trace could not be resolved, because no information about the language frames was collected.

See affected benchmarks

⚡ 23 improved benchmarks
❌ 17 regressed benchmarks
✅ 2368 untouched benchmarks
⏩ 12 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Walltime	threshold_neon[8]	1.4 µs	2.8 µs	-50.78%
❌	Walltime	compare_u8_neon	1.7 µs	3.1 µs	-43.67%
❌	Walltime	filter_neon[8]	967 ns	1,346 ns	-28.16%
❌	Walltime	filter_neon[64]	1.8 µs	2.4 µs	-24.02%
❌	Walltime	threshold_neon[64]	2.6 µs	3.4 µs	-23.01%
❌	Walltime	threshold_neon[80]	3.4 µs	4.4 µs	-22.55%
❌	Walltime	compare_int_neon	4.4 µs	5.3 µs	-16.49%
❌	Walltime	compare_int_constant_neon	3.9 µs	4.7 µs	-16.23%
❌	Walltime	compare_int_eq_neon	4.5 µs	5.3 µs	-15.59%
❌	CPU Simulation	filter_powerlaw_by_mostly_false[250000]	137.4 µs	158.1 µs	-13.08%
❌	Walltime	compare_int_nullable_neon	4.8 µs	5.6 µs	-12.95%
❌	Walltime	filter_neon[80]	2.5 µs	2.9 µs	-12.13%
❌	CPU Simulation	density_sweep_dense_runs[0.001]	52.3 µs	59 µs	-11.36%
❌	CPU Simulation	filter_random_by_mostly_false[250000]	162.5 µs	183.1 µs	-11.23%
❌	CPU Simulation	filter_random_by_power_law[250000]	165.6 µs	184.7 µs	-10.36%
❌	CPU Simulation	filter_powerlaw_by_power_law[250000]	165.8 µs	184.9 µs	-10.36%
❌	CPU Simulation	filter_powerlaw_by_random[250000]	152.5 µs	170 µs	-10.27%
⚡	Walltime	compare_f32_avx2	4.5 µs	2.4 µs	+84.35%
⚡	Walltime	compare_f32_avx512	3.4 µs	2.3 µs	+48.81%
⚡	Walltime	compare_u64_avx2	3.8 µs	2.8 µs	+36.99%
...	...	...	...	...	...

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/row-fn-cleanup/primitive-compare (393ac0c) with ct/row-fn-cleanup/fused-execution (69f5fc0)}

Footnotes

1. 12 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. ↩

[connortsui20]

Edit: reopened as a draft. The controlled measurements showed substantial AVX2 gains as well as ARM and AVX-512 regressions. Those results justify holding the PR for further work, but do not establish that the migration has no benefit. The earlier closure treated those two conclusions as equivalent.

#10741 isolates a useful float32 packing policy, and #10742 explores a smaller validity change. Neither replaces every benefit observed in the migration. Both remain independent drafts with their own measurement limits.

#10675 and this PR are being rebased while preserving their dependency. The saved measurements are historical evidence, not validation of the rebased heads.