EnglishPerformanceBenchmarks

Performance Benchmarks

Tracked Result Set

The repository now treats benchmark pages as summaries of tracked artifacts, not hand-maintained marketing tables.

  • Tracked JSON: benchmark/results/err091571-local-supported.json
  • Tracked report: benchmark/results/err091571-local-supported.md
  • Dataset manifest: benchmark/datasets.yaml

Verified Scope

  • Dataset: ENA ERR091571
  • Measured input: first 2,000 records streamed from read 1
  • Measured local tools: fqc, gzip, xz, bzip2
  • Deferred specialized peer: spring

This is a supported-local-toolset result, not a full-field FASTQ-compressor ranking.

Current Conclusion

For the currently tracked artifact:

  • fq-compressor does not lead the measured set on compression ratio
  • fq-compressor does not lead the measured set on compression speed
  • fq-compressor does not lead the measured set on decompression speed
  • comparison against specialized FASTQ peers remains incomplete because spring is still deferred

Why The Artifact Uses 2,000 Records

The tracked benchmark must be both public and reproducible.

The repository can stream a larger 20,000-record subset from ENA, but the current fq-compressor build regresses sharply on that exploratory size for this workload. The pinned artifact therefore uses a smaller public subset that still completes in the closeout validation loop.

Reproduce

./scripts/benchmark.sh \
  --dataset err091571-local-supported \
  --prepare \
  --build \
  --tools fqc,gzip,xz,bzip2,spring \
  --threads 1 \
  --runs 1

Scope Boundary

Do not read this page as proof of universal ranking. Unavailable or deferred specialized peers are outside the verified comparison scope until they are actually measured.

FQC FormatContributing