Skip to content

fix: pad delta remainder with last value - #9557

Open
rapour wants to merge 2 commits into
vortex-data:developfrom
rapour:fix/delta-span
Open

fix: pad delta remainder with last value#9557
rapour wants to merge 2 commits into
vortex-data:developfrom
rapour:fix/delta-span

Conversation

@rapour

@rapour rapour commented Aug 23, 2026

Copy link
Copy Markdown

Summary

Fixes #9556

Changes

Delta encoding remainder is padded with the last value instead of zeros.

Signed-off-by: rapour <reza.apour@gmail.com>
…ta is picked correctly

Signed-off-by: rapour <reza.apour@gmail.com>
@rapour

rapour commented Aug 24, 2026

Copy link
Copy Markdown
Author

Updated the following to make the CI green:

  • Linting issues addressed
  • The remainder is padded with the last value, so those positions hold real (if synthetic) data. The validity bitmap is transposed separately, and the bit transpose is not the same permutation as the integer transpose — so zero-filling the pad bits marks physical slots that hold real residuals as null. Bitpacking then skips patches for those values and decompression yields wrong results. Padding the mask with true keeps every real slot valid; the pad bits are sliced off by logical_len. Will be addressed fully in a separate PR.
  • Delta is now picked in scenarios where it would previously skipped so the snapshots are updated. There is only one case where compression is not improved which is compared to pco. This is due to the fact Delta scheme estimation does not include bases in the cost. Will be addressed in a follow-up PR.

@codspeed-hq

codspeed-hq Bot commented Aug 24, 2026

Copy link
Copy Markdown

Merging this PR will degrade performance by 10.85%

⚠️ 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.

❌ 1 regressed benchmark
✅ 1980 untouched benchmarks
⏩ 54 skipped benchmarks1

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

Mode Benchmark BASE HEAD Efficiency
Simulation compress_fsst[(500, 64, 8)] 530.5 µs 595 µs -10.85%

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.


Comparing rapour:fix/delta-span (0533a7a) with develop (e4b3421)

Open in CodSpeed

Footnotes

  1. 54 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.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

FastLanes delta padding could inflate max_minus_min span, skipping near-monotone columns

2 participants