Another collision detection discussion

[jonforum]

7 minutes ago, ivan.popelyshev said:

I remember back then i used box2d or box2d port with fluids, but of course now im doing it , but it was in 2014. I didnt even measure it back then, but yes, box2d was very performant

if your say it , i trust you, because your are Ivan, but i want just say the bench upper are not a good valide bench data ref!
we can't take those data for ref.

Edited November 12, 2019 by jonforum

[macshemp]

Hi @jonforum I didn't make that test but if you click on the links above it should show the same "tumbler test" using the different engines, you can then see the FPS for your machine. The hard coded FPS are those achieved by finscn on his machine, they aren't "dynamic"

 

@ivan.popelyshev I personally liked matterJS as it wasn't too old and had good documentation and was really easy to get running with PIXI. However as @mobileben has used both this and the typescript box2d and has demonstrated that box2sd is more accurate I am going to look into this in more detail before making a decision.

I think if you wanted something quick and easy matterJS would be acceptable but perhaps box2d might be more accurate in the long run. 

I think this is also a good contender: https://github.com/shakiba/planck.js 

It is a javascript rewrite of box2d and still has an active repository.

 

 

Edited November 12, 2019 by macshemp

[jonforum]

hum from individual test on Vivaldi browser (chromium) ~2sec test, ms per frames . (100% cpu usage for all!).
It a fast test, if you guys can also take and shares same kind of screen plz.

Hardware.

Result

link source code used: 
http://fatidol.com/phy-benchmark/
 

Edited November 12, 2019 by jonforum

[jonforum]

it interesting to see matter.js are the only one seem have a gpu compute, this is maybe good ?

Edit: oups no forget me!

Edited November 12, 2019 by jonforum

[tywang2006]

box2d is too heavy. I prefer lightweight as P2, 

in my opinion, at most case, P2 should be best option.