812201 - Scirra RenderPerf3 slower than chrome

Status: NEW

(Core :: JavaScript Engine, defect)

Description

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

The benchmark at http://www.scirra.com/labs/renderperf3/ gets around 142k in Chrome, vs. 66k in Firefox on my box with a 560 Ti.  On my work desktop with a faster CPU with a Quadro 2000, they both get around 33k.  No idea why Chrome gets such a big win on a faster GPU, but we should figure it out.

Comment 1

[Benoit Girard (:BenWa)]

Profile:
http://people.mozilla.com/~bgirard/cleopatra/#report=9ea4d029e60cbef674355a567cc83bfc71a91b05

View:
http://tinyurl.com/cf443cr

We spend about 71% of our time in JS. This benchmark would also improve slightly (~5% I'd guess) from using double buffering.

My guess is this benchmark needs some JS tuning.

Comment 2

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

Yep, looks like it.  Curious -- I wonder why that's different on my Quadro 2000 box?  I guess the 3D cost ends up overriding the JS cost?  Moving this to JS.  Unfortunately the code is obfuscated, we can probably reach out and get unobfuscated one.