scripts

README

StringZilla Scripts

This directory contains benchmarks, tests, and exploratory scripts for the StringZilla library, focused on internal functionality, rather than third-party alternatives.

• For comparative performance analysis, please refer to StringWars.
• To understand the distributional properties of hash functions, see HashEvals.

Benchmark Programs

Benchmarks validate SIMD-accelerated backends against serial baselines and measure throughput on real-world workloads.

• bench_find.cpp - bidirectional substring search, byte search, and byteset search
• bench_token.cpp - token-level operations: hashing, checksums, equality, and ordering
• bench_sequence.cpp - sorting, partitioning, and set intersections of string arrays
• bench_memory.cpp - memory operations: copies, moves, fills, and lookup table transformations
• bench_container.cpp - STL associative containers (std::map, std::unordered_map) with string keys
• bench_similarities.cpp - Levenshtein, Needleman-Wunsch, Smith-Waterman scoring on CPU
• bench_fingerprints.cpp - MinHash rolling fingerprints and multi-pattern search on CPU
• bench_similarities.cu - similarity scoring algorithms on CUDA GPUs
• bench_fingerprints.cu - fingerprinting algorithms on CUDA GPUs

All benchmarks support environment variables for configuration. Check file headers for details.

Test Programs

Unit tests validate correctness across all backends and programming languages.

• test_stringzilla.cpp - C++ API tests against STL baselines
• test_stringzilla.py - Python API tests against native strings
• test_stringzillas.cpp - parallel CPU backend tests
• test_stringzillas.cu - CUDA backend tests
• test.js - JavaScript API tests

Exploratory Notebooks

Jupyter notebooks for algorithm visualization and analysis.

• explore_levenshtein.ipynb - edit distance algorithms and diagonal traversal
• explore_fingerprint.ipynb - MinHash and rolling fingerprints
• explore_unicode.ipynb - UTF-8 handling and Unicode normalization