Spine performance

[Outfire]

I have canvas 1920 x 1080 and 5 different Spine animations that uses a lot of textures.
Add 20 animations of every type all over the screen

Test performance in two ways. For this I use GPU-Z and check GPU load parameter

1) Every animation has its own atlas with images and json file - GPU Load 47%
2) Every animation has its own json file but uses the same atlas with images - GPU Load 61%

Made the same tests with sprites as well,  total amount of sprites is 180, 10 of each type.  

1) Each sprite has its own download link - GPU Load 90 %
2) All textures are in the same spritesheet - GPU Load 63%

Why tests with Spine have opposite results?
 

PC config is Google Chrome v65, Windows 10, AMD Radeon HD 7450, Intel Xeon x5650

[ivan.popelyshev]

possibly because in case 1. FPS was less than 60. In case 2 there were less drawcalls and hotspot moved to GPU.

[Outfire]

Increase fps to 60 for spine test and now 1) 64% 2) 73%

For sprites test 1) 96% 2) 77%

[ivan.popelyshev]

I have no idea. Publish it.

[Outfire]

Here is the example.

For create single spritesheet i used unpacker for all spines atlas, took all images and pack them by texturepacker.

[HappinessSam]

I hope it's ok me piggybacking this thread for another, probably unrelated spine performance issue rather than starting another thread.

The project I'm working on has a lot of spine animation and also can have a very large canvas. I'm finding that if I have the game running then switch to another tab and wait more than 4-5 seconds, when I switch back to the project it freezes for a few seconds with the permorance analyser showing it being s. I have every spine with autoUpdate set to false and I have found that if I cap the delta time that can be passed into update it helps, but if with that there is still a noticable stutter.

To clarify, my update function looks something like this:

this.lastTime = this.lastTime || Date.now();
var timeDelta = (Date.now() - this.lastTime) * 0.001;
this.lastTime = Date.now();
if(timeDelta>0.1){
   timeDelta = 0.01;
}
spine.update(timeDelta);

Is this something anyone else has seen and is there anything that can be done to help?

[ivan.popelyshev]

and if you turn off debugger (just close the window), do you see thats stutter?

[HappinessSam]

It's still there, though nowhere near as bad - depending on the number of spine it's maybe half a second at most and usually just a quick stop stutter.

[Outfire]

@ivan.popelyshevHello again ?
I remembered my old test but on MacBook Pro (15-inch, 2018).
And the results.

For single atlas.

For multi atlas.

For windows, android and iOS it is ok, 5% difference,  but on mac time of using gpu x10. What could be the problem?

[ivan.popelyshev]

System and Drivers.

[Outfire]

On MacBook Pro (15-inch, 2017) same result, but on MacBook Pro (Retina, 15-inch, Mid 2015) everything is ok. ?
It's not a critical but i want to understand the problem.

[ivan.popelyshev]

/me dancing with tambourine like a shaman

try "transparent:true" in renderer options when you create it.

[Outfire]

Doesn't work. But i took my tambourine, dance with animation and found something strange.
If i use 25 "bear" or "elephant" animations, gpu need 2ms, but for 5 "bear" and 5 "elephant" - 10ms. ?