Go ECS Benchmarks

Comparative benchmarks for Go Entity Component System (ECS) implementations.

Disclaimer: This repository is maintained by the author of Ark.

Benchmark candidates

ECS	Tested	Latest	Activity
Ark	v0.8.3
Donburi	v1.15.7
go‑gameengine‑ecs	v0.9.0
GOKe	v3.1.0
unitoftime/ecs	v0.0.3
Volt	v1.7.0

Candidates are always displayed in alphabetical order.

In case you develop or use a Go ECS that is not in the list and that want to see here, please open an issue or make a pull request. See the section on Contributing for details.

In case you are a developer or user of an implementation included here, feel free to check the benchmarked code for any possible improvements. Open an issue if you want a version update.

Features

ECS	Type-safe API	ID-based API	Relations	Events[1]	Batches[2]	Command buffer
Ark	✅	✅	✅	✅	✅	❌
Donburi	✅	❌	❌	❌	❌	❌
go‑gameengine‑ecs	❌	✅	❌	❌	❌	❌
GOKe	✅	✅	❌	❌	✅	✅
unitoftime/ecs	✅	❌	❌	❌	❌	✅
Volt	✅	❌	❌	✅	❌	❌

[1] ECS lifecycle events, allowing to react to entity creation, component addition, ...
[2] Faster batch operations for entity creation etc.

Benchmarks

Last run: Fri, 21 Aug 2026 21:48:28 UTC
CPU: INTEL(R) XEON(R) PLATINUM 8573C

For each benchmark, the left plot panel and the table show the time spent per entity, while the right panel shows the total time.

Note that the Y axis has logarithmic scale in all plots. So doubled bar or line height is not doubled time!

All components used in the benchmarks have two float64 fields. The initial capacity of the world is set to 1024 where this is supported.

Query

N entities with components Position and Velocity. 10x N entities with components Position.

• Query all [Position, Velocity] entities, and add the velocity vector to the position vector.

N	Ark	Ark (tables)	Donburi	ggecs	GOKe	uot	Volt
1	59.79ns	59.16ns	51.44ns	36.76ns	11.56ns	10.08ns	62.92ns
4	16.72ns	14.97ns	23.86ns	10.80ns	3.46ns	4.03ns	16.46ns
16	4.76ns	4.02ns	17.53ns	4.72ns	1.26ns	2.42ns	4.86ns
64	2.60ns	1.52ns	16.23ns	3.53ns	0.76ns	2.15ns	2.19ns
256	1.26ns	0.75ns	15.56ns	3.02ns	0.54ns	2.27ns	1.36ns
1k	1.07ns	0.57ns	15.45ns	2.97ns	0.50ns	1.95ns	1.25ns
16k	1.09ns	0.70ns	17.60ns	3.20ns	0.68ns	1.93ns	1.13ns
256k	1.39ns	1.12ns	17.18ns	3.05ns	1.16ns	2.14ns	1.33ns
1M	1.52ns	1.08ns	25.83ns	2.94ns	1.17ns	2.17ns	1.38ns

Note: Donburi, unitoftime/ecs, and Volt use a callback-based approach for their query loops. As a result, iteration speed may degrade if the callback contains complex logic and the Go compiler is unable to inline it.

Query fragmented, inner

Query where the matching entities are fragmented over 32 archetypes.

N entities with components Position and Velocity. Each of these N entities has some combination of components C1, C2, ..., C5, so entities are fragmented over up to 32 archetypes.

• Query all [Position, Velocity] entities, and add the velocity vector to the position vector.

N	Ark	Ark (tables)	Donburi	GOKe	ggecs	uot	Volt
1	57.15ns	59.05ns	49.64ns	12.08ns	40.94ns	9.96ns	66.58ns
4	20.79ns	24.97ns	31.45ns	10.50ns	13.01ns	8.77ns	58.19ns
16	13.29ns	18.61ns	27.30ns	9.33ns	7.57ns	12.98ns	50.92ns
64	7.32ns	9.66ns	20.77ns	5.01ns	5.11ns	7.87ns	26.47ns
256	3.56ns	3.35ns	18.78ns	2.46ns	3.83ns	3.74ns	7.67ns
1k	1.90ns	1.33ns	15.97ns	1.03ns	3.28ns	2.37ns	3.32ns
16k	1.12ns	0.74ns	23.30ns	0.72ns	3.12ns	1.98ns	1.25ns
256k	1.44ns	1.12ns	26.54ns	1.21ns	2.89ns	2.17ns	1.41ns
1M	1.41ns	1.08ns	64.88ns	1.23ns	2.86ns	2.18ns	1.40ns

Note: Donburi, unitoftime/ecs, and Volt use a callback-based approach for their query loops. As a result, iteration speed may degrade if the callback contains complex logic and the Go compiler is unable to inline it.

Query fragmented, outer

Query where there are 256 non-matching archetypes.

N entities with components Position and Velocity. Another 4 * N entities with Position and some combination of 8 components C1, ..., C8, so these entities are fragmented over up to 256 archetypes.

• Query all [Position, Velocity] entities, and add the velocity vector to the position vector.

N	Ark	Ark (tables)	Donburi	ggecs	GOKe	uot	Volt
1	57.05ns	65.85ns	52.01ns	46.78ns	11.52ns	10.16ns	81.13ns
4	16.66ns	14.68ns	23.90ns	20.41ns	3.67ns	5.59ns	45.22ns
16	4.76ns	4.12ns	17.90ns	12.36ns	1.31ns	2.79ns	33.22ns
64	2.31ns	1.46ns	16.22ns	11.34ns	0.79ns	2.47ns	30.97ns
256	1.41ns	0.77ns	15.64ns	5.04ns	0.56ns	2.16ns	8.73ns
1k	1.06ns	0.58ns	17.50ns	3.47ns	0.50ns	1.95ns	2.97ns
16k	1.07ns	0.71ns	16.70ns	2.87ns	0.69ns	1.93ns	1.20ns
256k	1.49ns	1.09ns	17.53ns	2.88ns	1.17ns	2.28ns	1.33ns
1M	1.45ns	1.08ns	19.10ns	2.93ns	1.15ns	2.22ns	1.35ns

Note: Donburi, unitoftime/ecs, and Volt use a callback-based approach for their query loops. As a result, iteration speed may degrade if the callback contains complex logic and the Go compiler is unable to inline it.

Component random access

N entities with component Position. All entities are collected into a slice, and the slice is shuffled.

• Iterate the shuffled entities.
• For each entity, get its Position and sum up their X fields.

N	Ark	Donburi	GOKe	ggecs	uot	Volt
1	2.84ns	8.05ns	2.96ns	8.20ns	34.91ns	14.06ns
4	3.10ns	7.96ns	3.10ns	8.46ns	34.58ns	13.84ns
16	4.35ns	7.93ns	3.66ns	13.12ns	35.08ns	13.77ns
64	2.84ns	8.03ns	3.13ns	13.78ns	37.39ns	13.88ns
256	2.85ns	8.36ns	3.09ns	13.46ns	39.04ns	13.77ns
1k	2.93ns	11.45ns	3.27ns	16.78ns	40.81ns	13.76ns
16k	6.46ns	38.97ns	5.58ns	28.62ns	60.72ns	18.62ns
256k	9.08ns	180.70ns	14.22ns	117.61ns	176.19ns	35.91ns
1M	47.92ns	260.90ns	51.08ns	165.57ns	241.18ns	124.68ns

Create entities

• Create N entities with components Position and Velocity.

The operation is performed once before benchmarking, to exclude memory allocation, archetype creation etc. See the benchmark below for entity creation with allocation.

N	Ark	Ark (batch)	Donburi	ggecs	GOKe	uot	Volt
1	179.94ns	175.45ns	991.84ns	257.70ns	175.20ns	334.01ns	469.15ns
4	71.06ns	51.08ns	509.31ns	143.51ns	47.52ns	194.85ns	244.46ns
16	45.76ns	19.90ns	337.48ns	160.20ns	16.07ns	161.63ns	162.27ns
64	40.92ns	14.16ns	229.91ns	131.52ns	6.80ns	139.47ns	115.30ns
256	30.44ns	13.27ns	196.29ns	93.92ns	5.45ns	107.22ns	73.12ns
1k	29.76ns	12.15ns	182.12ns	90.58ns	4.63ns	273.57ns	76.12ns
16k	23.60ns	8.80ns	193.90ns	94.80ns	8.68ns	276.95ns	73.82ns
256k	25.19ns	8.72ns	194.94ns	196.40ns	10.81ns	463.05ns	72.84ns
1M	25.36ns	9.23ns	168.86ns	211.85ns	13.56ns	480.03ns	71.94ns

Create entities, allocating

• Create N entities with components Position and Velocity.

Each round is performed on a fresh world. This reflects the creation of the first entities with a certain components set in your game or application. As soon as things stabilize, the benchmarks for entity creation without allocation (above) apply.

Low N values might be biased by things like archetype creation and memory allocation, which is handled differently by different implementations.

N	Ark	Ark (batch)	Donburi	GOKe	ggecs	uot	Volt
1	9.18us	9.33us	5.84us	90.03ns	19.71us	3.44us	2.04us
4	2.36us	2.33us	1.72us	88.16ns	4.50us	1.19us	810.46ns
16	636.81ns	596.54ns	721.39ns	58.71ns	1.28us	475.51ns	329.60ns
64	186.66ns	153.98ns	395.64ns	65.85ns	498.43ns	267.52ns	182.96ns
256	70.56ns	53.45ns	285.43ns	52.63ns	315.50ns	200.83ns	125.45ns
1k	38.32ns	23.51ns	277.38ns	58.49ns	149.86ns	160.65ns	101.08ns
16k	48.66ns	39.29ns	401.29ns	68.05ns	233.62ns	182.10ns	123.20ns
256k	61.76ns	39.38ns	474.32ns	56.99ns	652.15ns	259.82ns	136.58ns
1M	51.18ns	36.64ns	433.97ns	56.68ns	2.00us	344.44ns	136.67ns

Create large entities

• Create N entities with 10 components C1, ..., C10.

The operation is performed once before benchmarking, to exclude things like archetype creation and memory allocation.

N	Ark	Ark (batch)	Donburi	ggecs	GOKe	uot	Volt
1	179.69ns	169.51ns	1.85us	346.60ns	182.97ns	518.08ns	2.16us
4	72.04ns	50.23ns	1.07us	237.84ns	51.14ns	442.95ns	1.66us
16	48.69ns	20.25ns	851.23ns	281.21ns	16.15ns	318.45ns	1.30us
64	41.05ns	13.76ns	613.56ns	159.59ns	6.84ns	291.21ns	894.73ns
256	39.23ns	11.84ns	539.85ns	135.60ns	5.87ns	205.10ns	817.21ns
1k	26.32ns	10.02ns	546.31ns	137.61ns	22.55ns	427.09ns	791.10ns
16k	24.49ns	8.79ns	621.91ns	146.35ns	22.14ns	530.53ns	796.29ns
256k	24.85ns	8.90ns	543.08ns	249.92ns	24.43ns	616.20ns	782.94ns
1M	24.93ns	9.27ns	623.29ns	261.40ns	28.69ns	629.36ns	791.34ns

Add/remove component

N entities with component Position.

• Add Velocity to all entities.
• Remove Velocity from all entities.

One iteration is performed before the benchmarking starts, to exclude memory allocation.

N	Ark	Ark (batch)	Donburi	ggecs	GOKe	uot	Volt
1	97.57ns	167.68ns	399.10ns	416.70ns	472.89ns	342.68ns	259.53ns
4	107.20ns	45.52ns	405.35ns	419.92ns	130.79ns	338.46ns	266.61ns
16	107.08ns	17.76ns	402.16ns	451.89ns	44.01ns	348.94ns	265.20ns
64	106.93ns	10.75ns	403.83ns	452.29ns	22.76ns	350.81ns	265.18ns
256	105.62ns	8.70ns	396.26ns	452.83ns	16.87ns	360.57ns	262.34ns
1k	104.14ns	8.70ns	400.32ns	467.81ns	17.14ns	733.56ns	261.68ns
16k	103.57ns	9.49ns	432.37ns	506.61ns	35.74ns	793.13ns	263.08ns
256k	104.54ns	10.61ns	405.65ns	820.63ns	44.41ns	1.24us	253.50ns
1M	103.61ns	11.74ns	407.66ns	939.90ns	46.10ns	1.41us	254.93ns

Add/remove component, large entity

N entities with component Position and 10 further components C1, ..., C10.

• Add Velocity to all entities.
• Remove Velocity from all entities.

One iteration is performed before the benchmarking starts, to exclude memory allocation.

N	Ark	Ark (batch)	Donburi	ggecs	GOKe	uot	Volt
1	337.06ns	440.33ns	1.11us	843.89ns	651.69ns	834.47ns	1.39us
4	385.52ns	122.62ns	1.14us	870.82ns	179.78ns	853.43ns	1.36us
16	389.82ns	47.93ns	1.12us	890.39ns	65.20ns	827.53ns	1.34us
64	392.16ns	25.45ns	1.11us	868.04ns	37.36ns	873.81ns	1.34us
256	381.17ns	22.10ns	1.10us	880.77ns	38.00ns	910.07ns	1.33us
1k	414.88ns	20.55ns	1.17us	898.10ns	89.89ns	1.47us	1.38us
16k	434.18ns	22.26ns	1.26us	1.09us	117.90ns	1.58us	1.54us
256k	538.60ns	45.43ns	1.33us	1.55us	138.98ns	2.03us	1.58us
1M	478.06ns	41.55ns	1.31us	1.63us	177.46ns	2.08us	1.59us

Delete entities

N entities with components Position and Velocity.

• Delete all entities

N	Ark	Ark (batch)	Donburi	GOKe	ggecs	uot	Volt
1	94.23ns	143.17ns	227.67ns	1.08us	411.38ns	109.61ns	254.69ns
4	47.85ns	39.70ns	93.18ns	259.69ns	194.07ns	56.98ns	142.49ns
16	47.68ns	14.29ns	67.87ns	78.04ns	147.26ns	47.63ns	96.05ns
64	33.66ns	6.55ns	45.00ns	25.42ns	100.31ns	38.18ns	62.08ns
256	25.31ns	4.82ns	36.54ns	13.78ns	77.28ns	32.80ns	44.18ns
1k	22.21ns	3.61ns	28.39ns	11.11ns	72.89ns	27.76ns	43.58ns
16k	15.67ns	3.19ns	27.98ns	11.98ns	72.83ns	28.95ns	33.78ns
256k	16.34ns	3.36ns	27.29ns	12.72ns	147.86ns	52.64ns	34.30ns
1M	16.73ns	3.50ns	28.10ns	13.60ns	275.72ns	74.35ns	35.64ns

Delete large entities

N entities with 10 components C1, ..., C10.

• Delete all entities

N	Ark	Ark (batch)	Donburi	ggecs	GOKe	uot	Volt
1	159.79ns	216.83ns	397.73ns	502.88ns	1.04us	107.37ns	669.10ns
4	125.20ns	62.50ns	215.90ns	307.89ns	268.46ns	52.73ns	480.52ns
16	114.93ns	23.00ns	132.91ns	231.65ns	90.06ns	49.28ns	287.82ns
64	77.14ns	10.79ns	98.10ns	146.53ns	33.43ns	35.30ns	213.86ns
256	77.26ns	8.11ns	84.74ns	127.62ns	18.50ns	31.10ns	189.54ns
1k	52.30ns	6.64ns	83.14ns	105.62ns	27.63ns	28.63ns	181.23ns
16k	49.20ns	7.62ns	83.70ns	117.51ns	21.30ns	26.48ns	180.49ns
256k	72.84ns	7.71ns	77.61ns	254.58ns	31.88ns	52.12ns	180.53ns
1M	79.52ns	15.11ns	148.99ns	446.70ns	56.77ns	68.81ns	182.33ns

Create world

• Create a new world

N	Ark	Donburi	GOKe	ggecs	uot	Volt
1	34.07us	3.51us	925.95us	276.03us	3.43us	23.36us

Popularity

Given that all tested projects are on Github, we can use the star history as a proxy here.

Running the benchmarks

Run the benchmarks using the following command:

go run . -test.benchtime=10x

On PowerShell use this instead:
go run . --% -test.benchtime=10x

The benchtime limit is required for some of the benchmarks that have a high setup cost which is not timed. They would take forever otherwise. The benchmarks can take up to one hour to complete.

To run a selection of benchmarks, add their names as arguments:

go run . query2comp query32arch

To create the plots, run plot/plot.py. The following packages are required:

• numpy
• pandas
• matplotlib

pip install -r ./plot/requirements.txt
python plot/plot.py

Contributing

Developers of ECS frameworks are welcome to add their implementation to the benchmarks. However, there are a few (quality) criteria that need to be fulfilled for inclusion:

• All benchmarks must be implemented, which means that the ECS must have the required features
• The ECS must be working properly and not exhibit serious flaws; it will undergo a basic review by maintainers
• The ECS must be sufficiently documented so that it can be used without reading the code
• There must be at least basic unit tests
• Unit tests must be run in the CI of the repository
• The ECS must not be tightly coupled to a particular game engine, particularly graphics stuff
• There must be tagged release versions; only tagged versions will be included here

Developers of included frameworks are encouraged to review the benchmarks, and to fix (or point to) misuse or potential optimizations.

Automated results PRs

Pull requests from the bench-results-update branch are opened automatically by the publish job in .github/workflows/benchmarks.yml, after a benchmark run on main — they only refresh README.md and docs/results/** (the plotted numbers/images), never benchmark code, and don't need the code-review criteria above. That branch is force-pushed on every run, so there's at most one such PR open at a time, reused rather than duplicated. Everything else (bench/**, go.mod, workflow, docs) is normal work, reviewed as usual.