Compare primitive values with RowFn

[connortsui20]

Summary

• Depends on Decode RowFn constants directly #9680

Moves primitive comparisons entirely onto RowFn to measure the limits of the generic path. This removes the handwritten SIMD and primitive-specific fallback.

Changes

PrimitiveCompare declares INFALLIBLE = true and routes every primitive type, comparison operator, and constant orientation through multiversioned RowVisitor::visit_bool. Cross-generated optimized x86-64 IR and assembly show that LLVM vectorizes the row comparisons, but this does not establish end-to-end benchmark parity. CodSpeed results for this PR cover the accumulated stack, including #9680 direct constant decoding.

[codspeed-hq]

Merging this PR will regress 24 benchmarks

⚠️ 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.

⚠️ 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

⚡ 83 improved benchmarks
❌ 24 regressed benchmarks
✅ 2032 untouched benchmarks
🆕 18 new benchmarks
⏩ 106 skipped benchmarks¹
🗄️ 4 archived benchmarks run²

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	WallTime	lt_i64_nullable_avx512	2.9 µs	4.4 µs	-35.26%
❌	WallTime	lt_i64_nullable_avx2	3 µs	4.4 µs	-32.55%
❌	WallTime	lt_i64_nullable_neon	3.5 µs	4.9 µs	-29.37%
❌	WallTime	compare_int_nullable_avx512	3.6 µs	5 µs	-28.24%
❌	WallTime	mul_u64_nonnull_neon	15.2 µs	20.7 µs	-26.72%
❌	WallTime	compare_int_nullable_avx2	3.9 µs	5.1 µs	-22.21%
❌	WallTime	compare_int_nullable_neon	5 µs	6.2 µs	-19.72%
❌	WallTime	eq_i64_constant_avx512	2.1 µs	2.6 µs	-17.39%
❌	Simulation	int_gt[16]	97.3 µs	116.8 µs	-16.68%
❌	Simulation	or_chain[16]	1.1 ms	1.4 ms	-16.08%
❌	Simulation	int_gt[1024]	103.2 µs	121.8 µs	-15.33%
❌	Simulation	or_chain[1024]	1.2 ms	1.4 ms	-15.12%
❌	WallTime	mul_i64_nonnull_neon	17.1 µs	20 µs	-14.6%
❌	WallTime	multiply_shapes_neon[(16384, PerRowPerRow)]	17.2 µs	20.1 µs	-14.57%
❌	Simulation	float_gt[16]	237.3 µs	277.7 µs	-14.54%
❌	Simulation	float_gt[1024]	262.3 µs	303.3 µs	-13.53%
❌	Simulation	is_not_null_pred[1024]	203.7 µs	234.9 µs	-13.27%
❌	Simulation	is_not_null_pred[16]	195 µs	224.4 µs	-13.09%
❌	Simulation	cold_misaligned[(16, 64)]	345.6 µs	391.4 µs	-11.7%
❌	Simulation	case_when_all_true[1000]	102.4 µs	115.4 µs	-11.32%
...	...	...	...	...	...

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/primitive-comparison-simd (139c28f) with develop (ba90f3d)³}

Footnotes

1. 106 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. ↩

2. 4 benchmarks were run, but are now archived. If they were deleted in another branch, consider rebasing to remove them from the report. Instead if they were added back, click here to restore them. ↩

3. No successful run was found on ct/row-fn-constant-decode (f21a243) during the generation of this report, so develop (ba90f3d) was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩