About performance of `indexBuffer` & `drawElements`.

recently , I try to create some examples about shader (just for studying), so I don’t use PIXI. I use a simple webgl wrapper igloojs at https://github.com/skeeto/igloojs

And I found a strange thing :
When I just render thousands of sprite (use same texture , so they’re batch) ---- no motion , no animation, no transform, no shader ,no filter ,no any effect, just “stand & still” , the igloojs is faster than pixi very much , very very very much , at least 3x. Even I try to use PIXI.glCore directly to render texture (similar to http://dev.goodboydigital.com/client/goodboy/million/ , but no physics, no particles shader), the igloojs is still faster very much.

So I read the code , I found the biggest difference is that igloojs use drawArrays(TRIANGLES...) , and no indexBuffer , I have to set the 6 vert manually.

And I found use drawArrays(gl.TRIANGLES...) is faster than gl.drawElements(gl.TRIANGLES...).

Is indexBuffer & drawElements the hotspot of performance ?

[finscn]

I refactor http://dev.goodboydigital.com/client/goodboy/million/ , create 2 new versions

with drawElements ( 25 fps on my MacBookPro 2015) http://fatidol.com/million-bunnies/draw-elements.html
(only support max 16384 sprites, Whatever set how many bunnies , it’s always less than 16384.)

with drawArrays ( 40 fps on my MacBookPro 2015) http://fatidol.com/million-bunnies/draw-arrays.html

You could find draw-arrays is more fast than draw-elements

[lock[bot]]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.