After merging your PR, Conbench analyzed the 4 benchmarking runs that have been run so far on merge-commit 6fe0ff0.

Benchmarks with performance regressions

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

All benchmark runs analyzed:

• Commit Run on arm64-t4g-2xlarge-linux at 2024-10-10 18:00:41Z
• Commit Run on amd64-c6a-4xlarge-linux at 2024-10-10 17:44:02Z
• Commit Run on test-mac-arm at 2024-10-10 18:02:45Z
• Commit Run on amd64-m5-4xlarge-linux at 2024-10-10 17:57:46Z

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 2024-10-10 17:44:02Z

  • approxEqualsBenchmark (Java) with source=java-micro, suite=arrow.vector.FloatingPointBenchmarks
  • BM_ReadInt64Column (C++) with params=<Repetition::OPTIONAL, Compression::LZ4>/1024/1024, source=cpp-micro, suite=parquet-column-io-benchmark
  • BM_ReadInt64Column (C++) with params=<Repetition::OPTIONAL, Compression::LZ4>/1024/16, source=cpp-micro, suite=parquet-column-io-benchmark
  • BM_ReadInt64Column (C++) with params=<Repetition::REPEATED, Compression::LZ4>/1024/16, source=cpp-micro, suite=parquet-column-io-benchmark

• Commit Run on test-mac-arm at 2024-10-10 18:02:45Z

  • FieldPathGetFromWideBatch (C++) with source=cpp-micro, suite=arrow-type-benchmark
  • ArrayScalarKernel (C++) with params=<DivideChecked, UInt64Type>/size:4194304/inverse_null_proportion:0, source=cpp-micro, suite=arrow-compute-scalar-arithmetic-benchmark
  • BM_ReadInt64Column (C++) with params=<Repetition::OPTIONAL, Compression::LZ4>/65536/1024, source=cpp-micro, suite=parquet-column-io-benchmark