Skip to content

Performance

UnitBuilds edited this page Aug 30, 2026 · 1 revision

Performance

Benchmarks

Run the built-in benchmark suite:

./target/release/velocity_mcp --benchmark

Protocol Parsing

Same tool call, full field extraction (apples-to-apples):

Operation Mean Latency Throughput
JSON-RPC parse + extract 722.6 ns ~1.38M req/s
NDA-native parse + Merkle + extract 459.1 ns ~2.18M req/s
NDA speedup 1.6x

Shared Memory Throughput

Operation Mean Latency Throughput
JSON-in-shmem R/W 38.8 ns 25.8M ops/s
NDA-native shmem R/W 12.6 ns 79.2M ops/s
NDA speedup 3.1x

Concurrent Scaling

Threads Throughput
1 2.86M req/s
4 7.01M req/s
8 12.7M req/s

vs Node.js MCP

Stdio, 200 requests per method (Intel Core i5-14400F):

Method Node.js VELOCITY-MCP Speedup
ping 0.573 ms 0.157 ms 3.6x
tools/list 1.050 ms 0.154 ms 6.8x
tools/call 0.546 ms 0.136 ms 4.0x
Overall 0.627 ms 0.164 ms 3.8x

Resource Usage

Metric VELOCITY-MCP Node.js MCP
Memory ~15 MB ~120 MB
Startup <50ms ~500ms

Criterion Benchmarks

For detailed statistical analysis:

cargo bench                    # All benchmarks
cargo bench --bench protocol_bench  # Protocol only
cargo bench -- --html          # With HTML report

See also: Transport-Modes, NDA-Format

Clone this wiki locally