lschmierer / ecs_bench Goto Github PK

Obviously, this is on the edge of the scope of this project. It would be nice to see how Rust implementations fare against more mature C++ ones on the same data sets.
Candidates (from another bench (cached)):

• EntityX
• Anax
• Artemis

No need for such a high precision here, it just distracts from the scale.

I'm interested in seeing memory consumption of the different frameworks. I'm curious how much overhead the data structures are adding. It seems for use cases other than sequential access, additional index structures would need to be added. Or e.g. for archetype iteration.

I haven't used an ECS in my own practice (still evaluating), so forgive my naivety if these overheads are not relevant.

While the benchmarks can show a rough difference, they're very far from real-world and they favor some approaches more than others I think. One point which should be improved for example is the "rendering" in the pos_vel benchmarks, because there's no code executed. I think it would also be great to have a benchmark with more systems than two.

I hope we can collect some ideas here and work out detailed specifications what the benchmarks should do. Additionally, we may consider removing libraries which aren't maintained anymore (except they're complete, of course).

Currently, in a pos-vel bench, we have N entities with both followed by M entities with only the position. This setup doesn't really favour any compromises other than simple linear vectors. I suggest inserting a pos-vel entity after each K pos-only entities (say, K=9).

This would highlight the differences between hash maps, vectors, and dense vectors.

Also maybe make some sort of table to put measurements into to compare.

...is it normal ?
https://github.com/lschmierer/ecs_bench/blob/master/benches/pos_vel_specs.rs#L45

• ecs
• specs
• recs
• trex
• constellation
• tinyecs