ZAP Protocol Benchmarks

Docs: ZAP benchmark suite · part of the ZAP Protocol

Real-world performance benchmarks for ZAP Protocol comparing against Protobuf, JSON, and traditional architectures.

Benchmark Categories

1. Serialization Benchmarks (bench-serialize/)

Core encoding/decoding performance comparisons.

2. Agent Benchmarks (bench-agents/)

MCP server memory overhead and multi-agent orchestration:

• Claude Code with 100 individual MCP servers
• Hanzo Dev with 1 ZAP router (proxying to 100 MCP servers)
• 20 parallel sub-agent task execution

3. Blockchain Benchmarks (bench-blockchain/)

Warp messaging and consensus operations:

• Cross-chain message encoding
• Validator set updates
• State proof verification
• Consensus round-trip times

4. Inference Benchmarks (bench-inference/)

Distributed AI inference:

• KV cache shard transfers
• Model weight distribution
• Batch prompt encoding

Quick Start

# Install dependencies
make setup

# Run all benchmarks
make bench

# Run specific benchmark suite
make bench-agents
make bench-blockchain
make bench-serialize
make bench-inference

# Generate reports
make report

Requirements

• Go 1.21+
• Python 3.11+
• Node.js 20+ (for MCP benchmarks)
• Docker (optional, for isolated tests)

Results

Results are written to results/ directory in JSON format and can be visualized with:

make charts

Methodology

All benchmarks follow these principles:

1. Reproducibility: Fixed seeds, controlled environments
2. Statistical rigor: Multiple iterations, percentile reporting
3. Fair comparison: Same data structures across formats
4. Real workloads: Based on actual Hanzo/Lux production patterns

See METHODOLOGY.md for detailed methodology.

License

Apache 2.0