cherrypick/benchmark_di/REPORT.md

Comparative DI Benchmark Report: cherrypick vs get_it vs riverpod vs kiwi

Benchmark Parameters

• chainCount = 100
• nestingDepth = 100
• repeat = 5
• warmup = 2

Benchmark Scenarios

1. RegisterSingleton — Registers and resolves a singleton. Baseline DI speed.
2. ChainSingleton — A dependency chain A → B → ... → N (singleton). Deep singleton chain resolution.
3. ChainFactory — All chain elements are factories. Stateless creation chain.
4. AsyncChain — Async chain (async factory). Performance on async graphs.
5. Named — Registers two bindings with names, resolves by name. Named lookup test.
6. Override — Registers a chain/alias in a child scope. Tests scope overrides.

---

Comparative Table: chainCount=100, nestingDepth=100, repeat=5, warmup=2 (Mean time, µs)

Scenario	cherrypick	get_it	riverpod	kiwi	yx_scope
chainSingleton	20.6	14.8	275.2	47.0	82.8
chainFactory	90.6	71.6	357.0	46.2	79.6
register	82.6	10.2	252.6	43.6	224.0
named	18.4	9.4	12.2	10.2	10.8
override	170.6	11.2	301.4	51.4	146.4
chainAsync	493.8	34.0	5,039.0	–	87.2

Peak Memory Usage (Peak RSS, Kb)

Scenario	cherrypick	get_it	riverpod	kiwi	yx_scope
chainSingleton	338,224	326,752	301,856	195,520	320,928
chainFactory	339,040	335,712	304,832	319,952	318,688
register	333,760	334,208	300,368	327,968	326,736
named	241,040	229,632	280,144	271,872	266,352
override	356,912	331,456	329,808	369,104	304,416
chainAsync	311,616	434,592	301,168	–	328,912

---

Analysis

• get_it remains the clear leader for both speed and memory usage (lowest latency across most scenarios; excellent memory efficiency even on deep chains).
• kiwi shows the lowest memory footprint in chainSingleton, but is unavailable for async chains.
• yx_scope demonstrates highly stable performance for both sync and async chains, often at the cost of higher memory usage, especially in the register/override scenarios.
• cherrypick comfortably beats riverpod, but is outperformed by get_it/kiwi/yx_scope, especially on async and heavy nested chains. It uses a bit less memory than yx_scope and kiwi, but can spike in memory/latency for override.
• riverpod is unsuitable for deep or async chains—latency and memory usage grow rapidly.
• Peak memory (RSS): usually around 320–340 MB for all DI; riverpod/kiwi occasionally drops below 300MB. named/factory scenarios use much less.
• Stability: yx_scope and get_it have the lowest latency spikes; cherrypick can show peaks on override/async; riverpod is least stable on async (stddev/mean much worse).

Recommendations

• get_it (and often kiwi, if you don't need async): best for ultra-fast deep graphs and minimum peak memory.
• yx_scope: best blend of performance and async stability; perfect for production mixed DI.
• cherrypick: great for modular/testable architectures, unless absolute peak is needed; lower memory than yx_scope in some scenarios.
• riverpod: only for shallow DI or UI wiring in Flutter.

---

Last updated: August 20, 2025. Please see scenario source for details.