How szmosi/python-speed-lab works
Python Speed Lab runs three benchmark categories—sum of squares, matrix multiplication, and Mandelbrot—across Python and accelerated implementations, records elapsed times, then displays and visualizes results. The sampled source shows the suite dispatching work and storing timings; the README also identifies CPU/GPU comparison and visualization as product goals. The Cython implementation is present in the tree but not sampled, so its internal behavior is not inferred.
Main entry points: benchmark_all.py Mandel benchmark · fast_code.pyx Cython methods
Components 15
Benchmark execution
- Benchmark suite
benchmark_all.py - Sum benchmark
benchmark_all.py - Matrix benchmark
benchmark_all.py - Mandel benchmark
benchmark_all.py
Compute methods
- Python methods
benchmark_all.py - Multiprocess methods
benchmark_all.py - NumPy methods
benchmark_all.py - Numba methods
benchmark_all.py - Cython methods
fast_code.pyx - GPU methods
benchmark_all.py
Results and visualization
- Timing results
benchmark_all.py - Result display
benchmark_all.py - Plots and animation
benchmark_all.py
Other components
- Engineer
- CUDA runtime
Connections 25
- Engineer → Benchmark suite: runs
benchmark_all.py - Benchmark suite → Sum benchmark: runs
benchmark_all.py - Benchmark suite → Matrix benchmark: runs
benchmark_all.py - Benchmark suite → Mandel benchmark: runs
benchmark_all.py - Sum benchmark → Python methods: benchmarks
benchmark_all.py - Sum benchmark → Multiprocess methods: benchmarks
benchmark_all.py - Sum benchmark → NumPy methods: benchmarks
benchmark_all.py - Sum benchmark → Numba methods: benchmarks
benchmark_all.py - Matrix benchmark → Python methods: benchmarks
benchmark_all.py - Matrix benchmark → Multiprocess methods: benchmarks
benchmark_all.py - Matrix benchmark → NumPy methods: benchmarks
benchmark_all.py - Matrix benchmark → Numba methods: benchmarks
benchmark_all.py - Matrix benchmark → GPU methods: benchmarks
benchmark_all.py - Mandel benchmark → Python methods: benchmarks
benchmark_all.py - Mandel benchmark → Multiprocess methods: benchmarks
benchmark_all.py - Mandel benchmark → NumPy methods: benchmarks
benchmark_all.py - Mandel benchmark → Numba methods: benchmarks
benchmark_all.py - Mandel benchmark → Cython methods: calls
benchmark_all.py - Mandel benchmark → GPU methods: benchmarks
benchmark_all.py - GPU methods → CUDA runtime: uses
benchmark_all.py - Sum benchmark → Timing results: records
benchmark_all.py - Matrix benchmark → Timing results: records
benchmark_all.py - Mandel benchmark → Timing results: records
benchmark_all.py - Benchmark suite → Result display: displays
benchmark_all.py - Benchmark suite → Plots and animation: plots
benchmark_all.py