After merging your PR, Conbench analyzed the 3 benchmarking runs that have been run so far on merge-commit 6cefbd3.

Benchmarks with performance regressions

There were 0 possible performance regressions, according to the lookback z-score method.

All benchmark runs analyzed:

• Commit Run on amd64-c6a-4xlarge-linux at 2026-04-06 16:24:31Z
• Commit Run on arm64-t4g-2xlarge-linux at 2026-04-06 18:40:03Z
• Commit Run on amd64-m5-4xlarge-linux at 2026-04-06 16:53:29Z

Unstable benchmarks with performance regressions

The following benchmark results indicate a possible performance regression, but are known to sometimes produce false positives when applying the lookback z-score analysis.

• Commit Run on amd64-c6a-4xlarge-linux at 2026-04-06 16:24:31Z

  • BM_UnpackUint16 (C++) with params=Sse42Unaligned/13/64, source=cpp-micro, suite=arrow-bpacking-benchmark

• Commit Run on arm64-t4g-2xlarge-linux at 2026-04-06 18:40:03Z

  • BM_DeltaBitPackingEncode_Int32_Narrow (C++) with params=65536, source=cpp-micro, suite=parquet-encoding-benchmark
  • BM_DecodeArrowBooleanPlain (C++) with params=DecodeArrowWithNull/num_values:65536/null_in_ten_thousand:1, source=cpp-micro, suite=parquet-encoding-benchmark
  • BM_DeltaLengthDecodingSpacedByteArray (C++) with params=max-string-length:64/batch-size:512, source=cpp-micro, suite=parquet-encoding-benchmark
  • BM_ByteStreamSplitDecode_Double_Neon (C++) with params=1024, source=cpp-micro, suite=parquet-encoding-benchmark
  • BM_DecodeArrowBooleanPlain (C++) with params=DecodeArrowWithNull/num_values:65536/null_in_ten_thousand:100, source=cpp-micro, suite=parquet-encoding-benchmark