guides/Router-Benchmarks.md

Router Benchmarks

Current benchmark conclusions should be read from practical API scenarios first (small and medium), with large treated as a stress/scaling check.

Last updated: 2026-02-19 03:17:48

Latest full run: benchmarks/results/bench_20260219_030929.csv

Runtime defaults: BENCH_ITERATIONS=5000, BENCH_WARMUP=500, BENCH_REPEATS=1

Core Strengths

Core has very strong warm routing throughput for practical route set sizes.
Core has the lowest memory footprint across scenarios in this benchmark harness.
Core remains stable in large warm mode while large one-shot cold mode is intentionally harsh.

Why Small/Medium Matter Most

Most production APIs do not rebuild and dispatch against a 10k+ route tree on every request.
large in this suite is primarily a stress test to expose scaling regressions.
For production readiness, prioritize small/medium trends and regressions.

Core Snapshot (Latest Full Run)

Benchmark	Elapsed (ms)	Ops/s	Memory
`core_loop:warm:small`	28,253	176 971,10	2,0 MB
`core_request:cold:small`	1 860,864	2 686,92	2,0 MB
`core_loop:warm:medium`	26,811	186 490,62	12,0 MB
`core_request:cold:medium`	18 736,168	266,86	12,0 MB
`core_loop:warm:large`	91,927	54 391,28	52,0 MB
`core_request:cold:large`	198 236,307	25,22	52,0 MB

Memory Context (Large Scenario)

Router	Warm Memory
`core_loop:warm:large`	52,0 MB
`laravel:warm:large`	110,0 MB
`symfony:warm:large`	142,0 MB
`fastroute:warm:large`	144,0 MB
`slim:warm:large`	146,0 MB

Methodology Notes

Core has two relevant modes:
core_loop:warm:*: compiled loop-style dispatch
core_request:cold:*: one-shot full rebuild + dispatch (harsh by design)
Laravel warm setup uses compiled routes (route:cache equivalent) in this harness.
Results are single-process microbenchmarks; compare trends over multiple runs for confidence.

Regenerate

Run full benchmark:

php benchmarks/bin/benchmark_router.php

Run focused benchmark (example):

BENCH_FRAMEWORKS=core BENCH_SCENARIOS=small,medium BENCH_MODES=warm php benchmarks/bin/benchmark_router.php