Function benchmark_memory

benchmark/memory_benchmark.py:54–109

Benchmark memory usage for a dataset.

(name: str, data: dict)

def benchmark_memory(name: str, data: dict):
    """Benchmark memory usage for a dataset."""
    print(f"\n{'='*60}")
    print(f"Memory Benchmark: {name}")
    print(f"{'='*60}")

    # Python object memory
    obj_memory = get_memory_size(data)
    print(f"\n📦 Python Object:")
    print(f"  Memory: {format_size(obj_memory):>10} ({obj_memory:,} bytes)")

    # JSON string memory
    json_str = json.dumps(data, indent=2)
    json_str_memory = sys.getsizeof(json_str)
    json_bytes = json_str.encode('utf-8')
    json_bytes_memory = len(json_bytes)

    print(f"\n📄 JSON Format:")
    print(f"  String object memory: {format_size(json_str_memory):>10} ({json_str_memory:,} bytes)")
    print(f"  UTF-8 bytes size:     {format_size(json_bytes_memory):>10} ({json_bytes_memory:,} bytes)")

    # TOON string memory
    toon_str = encode(data)
    toon_str_memory = sys.getsizeof(toon_str)
    toon_bytes = toon_str.encode('utf-8')
    toon_bytes_memory = len(toon_bytes)

    print(f"\n🎯 TOON Format:")
    print(f"  String object memory: {format_size(toon_str_memory):>10} ({toon_str_memory:,} bytes)")
    print(f"  UTF-8 bytes size:     {format_size(toon_bytes_memory):>10} ({toon_bytes_memory:,} bytes)")

    # Calculate savings
    string_savings = ((json_str_memory - toon_str_memory) / json_str_memory) * 100
    bytes_savings = ((json_bytes_memory - toon_bytes_memory) / json_bytes_memory) * 100

    print(f"\n💾 Memory Savings:")
    print(f"  String memory: {string_savings:.1f}% smaller")
    print(f"  Bytes size:    {bytes_savings:.1f}% smaller")

    # Practical impact
    print(f"\n💡 Practical Impact:")
    print(f"  If you send this data to an LLM API:")
    print(f"    • JSON uses {json_bytes_memory:,} bytes of network bandwidth")
    print(f"    • TOON uses {toon_bytes_memory:,} bytes of network bandwidth")
    print(f"    • You save {json_bytes_memory - toon_bytes_memory:,} bytes per request!")

    return {
        'name': name,
        'obj_memory': obj_memory,
        'json_str_memory': json_str_memory,
        'json_bytes_memory': json_bytes_memory,
        'toon_str_memory': toon_str_memory,
        'toon_bytes_memory': toon_bytes_memory,
        'string_savings': string_savings,
        'bytes_savings': bytes_savings,
    }