Performance and Benchmarks

Benchmarks run from the v2 package directory. Use quick, full, and stress profiles to control coverage.

cd dictsqlite_v2/dictsqlite
python benchmark/benchmark_all.py --profile quick
python benchmark/analyze_results.py

Measured axes

• Data size
• Record count
• Batch size
• Storage and persistence modes
• Cold reads, hot writes, delete, clear, and threaded access

GitHub Actions

The automatic and manual performance workflows produce CSV, images, GitHub benchmark JSON, and Markdown in the Actions Summary. Comparison is informational to avoid false alerts from runner variance.