barbell 🏋️

Benchmark suite for comparing language performance across Hemlock, HemlockC, C, Python, and JavaScript.

Benchmarks

Micro-benchmarks

• fib - Recursive Fibonacci (CPU-bound, function call overhead)
• array_sum - Sum large array (memory access, iteration)
• string_concat - String concatenation (allocation, GC pressure)
• primes_sieve - Sieve of Eratosthenes (array access, loops)
• hash_table - Hash table insert/lookup/delete (dictionary operations)

Algorithms

• quicksort - Quicksort algorithm (recursion, array manipulation)
• merge_sort - Merge sort algorithm (recursion, array copying)
• binary_tree - Binary search tree insert and traversal (object allocation, recursion)
• graph_bfs - Breadth-first search on graph (queue operations, adjacency list traversal)
• matrix_multiply - Matrix multiplication (2D arrays, nested loops, floating point)

Real-world

• json_serialize - JSON serialization (object creation, string building)
• json_deserialize - JSON parsing (string parsing, object creation)
• hash_sha256 - SHA-256 hashing (bitwise operations, loops, array manipulation)
• http_test - HTTP request/response building and parsing (string operations)

Usage

./run.sh              # Run all benchmarks
./run.sh fib          # Run specific benchmark
./run.sh --quick      # Quick mode (smaller inputs)

Results

Results are output as JSON for easy comparison and visualization.

Adding a benchmark

1. Create implementations in benchmarks/<name>/ for each language
2. Add benchmark config to benchmarks.json
3. Run ./run.sh <name> to test