Add checked arithmetic compute

[gatesn]

FLUPs:

• Arithmetic array to invert lhs/rhs for constant checks
• Special handling for vectors with many nulls (iterate only the sparse non-null indices)

[codspeed-hq]

CodSpeed Performance Report

Merging #5096 will degrade performances by 10.35%

_{Comparing ngates/arithmetic-buffer-compute (d042d45) with develop (60ce4b5)¹}

Summary

⚡ 90 improvements
❌ 3 regressions
✅ 1225 untouched
🆕 7 new
⏩ 29 skipped²

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Benchmarks breakdown

	Benchmark	BASE	HEAD	Change
⚡	chunked_dict_primitive_canonical_into[f32, (1000, 10, 10)]	107.8 µs	92.5 µs	+16.45%
⚡	chunked_dict_primitive_canonical_into[f32, (1000, 10, 100)]	837.1 µs	715.7 µs	+16.97%
⚡	chunked_dict_primitive_canonical_into[f32, (1000, 100, 10)]	106.6 µs	91.5 µs	+16.59%
⚡	chunked_dict_primitive_canonical_into[f32, (1000, 100, 100)]	840.7 µs	721.1 µs	+16.59%
⚡	chunked_dict_primitive_canonical_into[f32, (1000, 1000, 10)]	127.9 µs	112.4 µs	+13.8%
⚡	chunked_dict_primitive_canonical_into[f32, (1000, 1000, 100)]	1,023.2 µs	900.2 µs	+13.67%
⚡	chunked_dict_primitive_canonical_into[f64, (1000, 10, 10)]	136.1 µs	120.3 µs	+13.15%
⚡	chunked_dict_primitive_canonical_into[f64, (1000, 10, 100)]	1,071.3 µs	947.2 µs	+13.1%
⚡	chunked_dict_primitive_canonical_into[f64, (1000, 100, 10)]	139.6 µs	123.9 µs	+12.7%
⚡	chunked_dict_primitive_canonical_into[f64, (1000, 100, 100)]	1.1 ms	1 ms	+12.42%
⚡	chunked_dict_primitive_canonical_into[f64, (1000, 1000, 10)]	172.1 µs	156.2 µs	+10.2%
⚡	chunked_dict_primitive_canonical_into[u32, (1000, 10, 10)]	108.3 µs	92.9 µs	+16.62%
⚡	chunked_dict_primitive_canonical_into[u32, (1000, 10, 100)]	840.7 µs	718.1 µs	+17.06%
⚡	chunked_dict_primitive_canonical_into[u32, (1000, 100, 10)]	109.6 µs	94.1 µs	+16.42%
⚡	chunked_dict_primitive_canonical_into[u32, (1000, 100, 100)]	845.1 µs	724.1 µs	+16.7%
⚡	chunked_dict_primitive_canonical_into[u32, (1000, 1000, 10)]	128.6 µs	112.9 µs	+13.84%
⚡	chunked_dict_primitive_canonical_into[u32, (1000, 1000, 100)]	1,004.3 µs	880.1 µs	+14.12%
⚡	chunked_dict_primitive_canonical_into[u64, (1000, 10, 10)]	135.9 µs	120.2 µs	+13.1%
⚡	chunked_dict_primitive_canonical_into[u64, (1000, 10, 100)]	1,105 µs	978 µs	+12.99%
⚡	chunked_dict_primitive_canonical_into[u64, (1000, 100, 10)]	137.7 µs	122.1 µs	+12.78%
...	...	...	...	...

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

Footnotes

1. No successful run was found on develop (d4bab58) during the generation of this report, so 60ce4b5 was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩

2. 29 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. ↩

[codecov]

Codecov Report

❌ Patch coverage is 65.99462% with 253 lines in your changes missing coverage. Please review.
✅ Project coverage is 85.58%. Comparing base (60ce4b5) to head (d042d45).
⚠️ Report is 4 commits behind head on develop.

Files with missing lines	Patch %	Lines
vortex-array/src/compute/arrays/arithmetic.rs	44.31%	142 Missing ⚠️
vortex-compute/src/arithmetic/mod.rs	50.00%	21 Missing ⚠️
vortex-compute/src/arithmetic/buffer_checked.rs	88.72%	15 Missing ⚠️
vortex-vector/src/primitive/vector.rs	0.00%	14 Missing ⚠️
vortex-vector/src/primitive/vector_mut.rs	0.00%	14 Missing ⚠️
vortex-compute/src/arithmetic/pvector.rs	84.81%	12 Missing ⚠️
vortex-compute/src/arithmetic/buffer.rs	84.37%	10 Missing ⚠️
vortex-array/src/array/operator.rs	46.15%	7 Missing ⚠️
vortex-compute/src/logical/and.rs	0.00%	6 Missing ⚠️
vortex-compute/src/logical/or.rs	0.00%	6 Missing ⚠️
... and 2 more

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[connortsui20]

This is just a question: how do we handle add where values can be null currently? Do we even need to worry about this if we are just taking the OR of the masks?

[connortsui20]

I'm not sure I understand the purpose of this? Could you instead just implement From<ArithmeticOperator> for ArithmeticEncoding?

[connortsui20]

Also hash the encoding since it contains which operation?

[connortsui20]

LGTM