@pokertools/bench

Benchmark suites for the evaluator plus load/soak scripts for the API, workers and sockets.

Usage

npm run bench                          # full suite (requires API + Redis for API benches)
npm run bench -w @pokertools/bench -- evaluator   # evaluator benchmarks only

Benchmark output is benchmark.js-style table: ops/sec, margin of error and the slowest/fastest ratio per suite.

Prerequisites

Suite	Needs API running	Needs Redis
Evaluator	❌	❌
API	✅	✅
Workers	❌	✅
Socket	✅	✅

Start the stack with npm run dev:api + npm run dev:workers (or the Docker stack, see Deployment).

What is benchmarked

Suite	Measures
Evaluator	5/6/7-card evaluation throughput (hands/sec), string vs integer API
API	Request latency/throughput against Fastify routes
Workers	BullMQ queue throughput (settle, archive, deal)
Socket	WebSocket session churn and state-update fan-out

Evaluator comparison

The evaluator suite compares @pokertools/evaluator against popular JS libraries (phe, poker-evaluator, pokersolver):

Implementation	Approx. throughput
@pokertools/evaluator (integer API)	millions of hands/sec
phe	fast C-style tables, lower than evaluator
poker-evaluator	moderate
pokersolver	slowest (high-level string API)

TIP

Published throughput figures are illustrative of the local machine — run the suite on your own hardware for numbers you can rely on.

Writing a new benchmark

import Benchmark from "benchmark";
import { evaluateStrings } from "@pokertools/evaluator";

const hand = ["As", "Ks", "Qs", "Js", "Ts"];

new Benchmark("evaluate royal flush", {
  fn: () => evaluateStrings(hand),
  onComplete: (event: any) => console.log(String(event.target)),
}).run();

Load & soak scripts

• api/* — sustained request load against a running API instance
• workers/* — queue saturation tests
• socket/* — connection churn and fan-out soak

These require the API (and Redis) to be running — see the API docs.