How guy5116/speed-test works
Speed-Test is a command-line benchmark suite comparing implementations of six workloads across fifteen languages, plus NumPy for three vectorizable workloads. The principal workflow is to choose a scale and run count, execute the available implementations on matching inputs, validate checksums, and present per-language timings, ratios, summaries, and human-scale projections. The README documents the workload and comparison stages; sampled source confirms NumPy’s vectorized kernels and run.py’s timing cache and result handling. The excerpts do not expose all runner dispatch details, so some documented links use the README as evidence.
Main entry points: run.py Run coordinator · bench.c Language implementations · bench_numpy.py NumPy kernels
Components 15
Benchmark execution
- Run coordinator
run.py - Language implementations
bench.c - NumPy kernels
bench_numpy.py
Workload kernels
- Mandelbrot
bench.c - Sieve
bench.c - Quicksort
bench.c - Word count
bench.c - Binary trees
bench.c - Matrix multiply
bench.c
Comparison and results
- Checksum validation
run.py - Timing comparison
run.py - Results report
run.py
Timing support
- Progress estimates
run.py - Timing cache
run.py
Other components
- CLI user
Connections 17
- CLI user → Run coordinator: sets options
README.md - Run coordinator → Language implementations: runs implementations
README.md - Run coordinator → NumPy kernels: runs eligible kernels
README.md - Language implementations → Mandelbrot: implements
README.md - Language implementations → Sieve: implements
README.md - Language implementations → Quicksort: implements
README.md - Language implementations → Word count: implements
README.md - Language implementations → Binary trees: implements
README.md - Language implementations → Matrix multiply: implements
README.md - NumPy kernels → Mandelbrot: vectorizes
bench_numpy.py - NumPy kernels → Sieve: vectorizes
bench_numpy.py - NumPy kernels → Matrix multiply: vectorizes
bench_numpy.py - Run coordinator → Checksum validation: validates checksums
README.md - Run coordinator → Timing comparison: collects timings
README.md - Timing comparison → Results report: renders results
README.md - Run coordinator → Progress estimates: estimates remaining time
README.md - Run coordinator → Timing cache: loads and saves
run.py