Compare primitive values with RowFn

[connortsui20]

Depends on #10014.

Summary

• Tracking Issue: Epic: Row-oriented scalar functions #9128
• Replaces Compare primitive values with RowFn #9703 under the renamed branch. The original discussion remains there.

Moves primitive comparisons onto RowFn so decoding, constants, validity, and packed Boolean output use the shared row executor.

Changes

PrimitiveCompare dispatches primitive types and comparison operators through multiversioned RowVisitor::visit_bool, replacing the primitive lane kernels and their operand wrapper. Float comparisons retain Vortex's total ordering.

Depends on #10014 for execution-allocator ownership of Boolean output. The comparison tests added in #9948 are retained and now exercise the RowFn implementation.

Validation before the rebase: 168 focused comparison, mask, and RowFn tests passed on the combined stack, including comparison_uses_execution_allocator, along with cargo clippy -p vortex-array --all-targets --all-features -- -D warnings. Tests, formatting, and benchmarks were not rerun locally after the rebase.

[codspeed]

Merging this PR will degrade performance by 5.81%

⚠️ Unknown Walltime execution environment detected

Using the Walltime instrument on standard Hosted Runners will lead to inconsistent data.

For the most accurate results, we recommend using CodSpeed Macro Runners: bare-metal machines fine-tuned for performance measurement consistency.

⚠️ 12 benchmarks spent significant time in system calls

System calls cannot be consistently instrumented, so they are not included in the measure, which understates the real cost. Please switch to the Walltime instrument to accurately measure system calls.

Measurement and system calls

• decode_primitives[i64, (1000, 32)]
• random_i128[0.01]
• chunked_dict_primitive_into_canonical[f64, (1000, 1000, 10)]
• density_sweep_dense_runs[0.001]
• decompress[u32, (1000, 256)]
• decompress[u16, (4000, 256)]
• take_varbin
• filter_run_end[len=4096_run=16_density=0.1]
• zstd_compress_string
• pcodec_compress_f64
• …and 2 more

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

⚡ 12 improved benchmarks
❌ 33 regressed benchmarks
✅ 2138 untouched benchmarks
⏩ 359 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Simulation	filter_powerlaw_by_mostly_true[250000]	103.3 µs	149 µs	-30.67%
❌	WallTime	compare_u8_avx512	1.5 µs	1.9 µs	-22.19%
❌	Simulation	take[routing/same_chunk/primitive/nonnull/chunks=32/indices=1000]	181.6 µs	224 µs	-18.92%
❌	WallTime	compare_u8_neon	1.9 µs	2.3 µs	-18.34%
❌	Simulation	int_gt[16]	94.6 µs	115 µs	-17.72%
❌	WallTime	compare_u8_avx2	1.6 µs	1.9 µs	-16.51%
❌	WallTime	lt_i64_nullable_avx512	2.6 µs	3.1 µs	-16.31%
❌	Simulation	int_gt[1024]	99.8 µs	118.7 µs	-15.92%
❌	Simulation	or_chain[16]	308.2 µs	365.5 µs	-15.68%
❌	Simulation	float_gt[16]	215.2 µs	252.3 µs	-14.73%
❌	Simulation	bench_compare_primitive[(10000, 2)]	65.3 µs	76.4 µs	-14.53%
❌	Simulation	or_chain[1024]	321.3 µs	375 µs	-14.32%
❌	Simulation	float_gt[1024]	240.2 µs	280.1 µs	-14.22%
❌	Simulation	i64_random[256]	4.3 ms	5 ms	-14.11%
❌	Simulation	i64_lt[true]	80.8 µs	93.7 µs	-13.8%
❌	Simulation	i64_lt[false]	80.7 µs	93.6 µs	-13.76%
❌	Simulation	i64_random_chunked[256]	15.9 ms	18.3 ms	-13.51%
❌	Simulation	baseline_lt[4, 1024]	70.7 µs	81.7 µs	-13.44%
❌	Simulation	bench_compare_sliced_dict_primitive[(3333, 10000)]	74 µs	84.8 µs	-12.69%
❌	Simulation	is_not_null_pred[1024]	188.8 µs	216.2 µs	-12.66%
...	...	...	...	...	...

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-compare-primitive (32ce8ba) with develop (c1043b3)}

Footnotes

1. 359 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. ↩

[github-actions]

This PR has been marked as stale because it has been open for 14 days with no activity. Please comment or remove the stale label if you wish to keep it active, otherwise it will be closed in 7 days