[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Public WebGL] GL_RENDERER string needed for performant apps



I support this proposal.

There are additional benefits I see in providing unmasked renderer, vendor and translated shader source.

GPU vendor/make could be represented in statistics, giving WebGL users a better picture what the hardware landscape of their potential users looks like. I've recently started to write an update to webglstats (preview there http://webglstats.com/v2/) and would gladly add this information.

Having the renderer/vendor available allows one to more accurately pinpoint issues not just with performance, but also with failing shaders or use use of features. This is a great boon when handling user support requests, and can provide valuable information from the field to browser vendors where additional work on improving drivers or implementations is required.

The translated shader source can aide with the aforementioned point as well, and it can give developers the opportunity to see if their shader has been translated in an "odd" way (like unrolling a 1000 iteration loop) and react to that case by providing a different shader source.

Lastly it may ease some pressure off the browsers white/black lists, as services could be developed that implement such logic on an application level, giving WebGL developers the opportunity to choose a "level" they are comfortable with (very simple use of WebGL will not require as many blacklisted devices as uses that are both feature and performance intensive).