After merging your PR, Conbench analyzed the 7 benchmarking runs that have been run so far on merge-commit 074d45f.

Benchmarks with performance regressions

There were 4 possible performance regressions, according to the lookback z-score method.

Benchmarks with regressions:

• Commit Run on ursa-i9-9960x at 2024-04-05 09:14:51Z
  • dataset-selectivity (Python) with dataset=nyctaxi_multi_ipc_s3, selectivity=10%
  • dataset-selectivity (Python) with dataset=nyctaxi_multi_parquet_s3, selectivity=10%
  • dataset-selectivity (Python) with dataset=nyctaxi_multi_ipc_s3, selectivity=1%
  • dataset-selectivity (Python) with dataset=nyctaxi_multi_parquet_s3, selectivity=100%

All benchmark runs analyzed:

• Commit Run on ec2-t3-xlarge-us-east-2 at 2024-04-05 09:13:56Z
• Commit Run on test-mac-arm at 2024-04-05 10:09:01Z
• Commit Run on ec2-m5-4xlarge-us-east-2 at 2024-04-05 09:46:02Z
• Commit Run on ursa-i9-9960x at 2024-04-05 09:14:51Z
• Commit Run on arm64-t4g-linux-compute at 2024-04-05 10:37:02Z
• Commit Run on ursa-thinkcentre-m75q at 2024-04-05 12:53:28Z
• Commit Run on arm64-m6g-linux-compute at 2024-04-05 10:32:03Z

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 test-mac-arm at 2024-04-05 10:09:01Z

  • BM_PlainDecodingSpacedByteArray (C++) with params=max-string-length:64/batch-size:2048, source=cpp-micro, suite=parquet-encoding-benchmark
  • MoveShortVector (C++) with params=<STATIC_VECTOR(std::string)>, source=cpp-micro, suite=arrow-small-vector-benchmark

• Commit Run on arm64-t4g-linux-compute at 2024-04-05 10:37:02Z

  • BM_DefinitionLevelsToBitmapRepeatedMostPresent (C++) with source=cpp-micro, suite=parquet-level-conversion-benchmark

• Commit Run on arm64-m6g-linux-compute at 2024-04-05 10:32:03Z

  • IntegerFormatting (C++) with params=, source=cpp-micro, suite=arrow-value-parsing-benchmark