Improve compact compression choices

[joseph-isaacs]

Summary

Improve the compact BtrBlocks compression model and scheme set used to produce the files explored by the follow-up Vortex Web compression diff viewer.

Changes

• Account for ALP exception patch structure and index width when estimating encoded size.
• Add integer and float Zstd schemes to the compact compressor.
• Refine bitpacking and ALP patch-selection costs.
• Share patch-size calculations across schemes.
• Increase the value framing used by compact Zstd string and binary compression.
• Add focused tests for the updated estimates and scheme choices.

Stack

1. This PR: compression/array-layer choices.
2. Follow-up: Vortex Web compression diff viewer (Add compression diff viewer to Vortex Web #9603).

[codspeed-hq]

Merging this PR will degrade performance by 88.38%

⚠️ 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

⚡ 3 improved benchmarks
❌ 39 regressed benchmarks
✅ 2055 untouched benchmarks
⏩ 106 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Simulation	alp_compress_f64	171.2 µs	7,487.4 µs	-97.71%
❌	Simulation	compress_alp[f64, (10000, 0.0, 1.0)]	310.4 µs	10,081.1 µs	-96.92%
❌	Simulation	compress_alp[f64, (10000, 0.0, 0.95)]	319.5 µs	10,090.2 µs	-96.83%
❌	Simulation	compress_alp[f64, (10000, 0.0, 0.25)]	319.9 µs	10,090.5 µs	-96.83%
❌	Simulation	compress_alp[f64, (10000, 0.01, 1.0)]	539.8 µs	10,310.2 µs	-94.76%
❌	Simulation	compress_alp[f64, (10000, 0.01, 0.25)]	563.8 µs	10,333.6 µs	-94.54%
❌	Simulation	compress_alp[f64, (10000, 0.01, 0.95)]	568.6 µs	10,338.4 µs	-94.5%
❌	Simulation	compress_alp[f64, (10000, 0.1, 1.0)]	574.2 µs	10,329.8 µs	-94.44%
❌	Simulation	compress_alp[f64, (1000, 0.0, 1.0)]	105.4 µs	1,815.4 µs	-94.19%
❌	Simulation	compress_alp[f64, (10000, 0.1, 0.25)]	653.1 µs	10,408.5 µs	-93.73%
❌	Simulation	compress_alp[f64, (1000, 0.0, 0.25)]	115.2 µs	1,825.1 µs	-93.69%
❌	Simulation	compress_alp[f64, (1000, 0.0, 0.95)]	115.2 µs	1,825.1 µs	-93.69%
❌	Simulation	compress_alp[f64, (10000, 0.1, 0.95)]	666.8 µs	10,422.4 µs	-93.6%
❌	Simulation	compress_alp[f32, (10000, 0.0, 1.0)]	180 µs	2,671 µs	-93.26%
❌	Simulation	compress_alp[f32, (10000, 0.0, 0.25)]	190.5 µs	2,681.3 µs	-92.89%
❌	Simulation	compress_alp[f32, (10000, 0.0, 0.95)]	190.5 µs	2,681.3 µs	-92.89%
❌	Simulation	compress_alp[f64, (1000, 0.01, 1.0)]	145.4 µs	1,855.4 µs	-92.16%
❌	Simulation	compress_alp[f64, (1000, 0.1, 1.0)]	151.6 µs	1,858.3 µs	-91.84%
❌	Simulation	compress_alp[f64, (1000, 0.01, 0.25)]	157.8 µs	1,867.8 µs	-91.55%
❌	Simulation	compress_alp[f64, (1000, 0.01, 0.95)]	161.2 µs	1,871.2 µs	-91.39%
...	...	...	...	...	...

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 ji/compression-model-and-patches (847e7b5) with develop (17bd9e2)}

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. ↩