News for hackers. Stuff that matters.
Re:Introducing Paper by RyanZAG on Thursday January 30, 2014 @07:41AM (#7150299)
I'm impressed by how much CPU this website is able to use. It pegged all 8 cores at 100% while I had the website open on firefox/win7
Re:Introducing Paper by Cthulhu_ on Thursday January 30, 2014 @08:21AM (#7150487)
And yet, pages / applications like this using a lot of CPU should be plenty of incentive and test cases for both web developers and browser developers to optimize their respective applications. Like how 3DMark and high-end video games like (at the time) Crysis push the GPU manufacturers and driver programmers to achieve more performance in those respective applications for its PR value.

'50% less CPU usage on the Facebook Paper page' is a nice line in a browser's changelog.

Re:Introducing Paper by weddpros on Thursday January 30, 2014 @09:00AM (#7150672)
Looks a lot like a bug somewhere in Chrome...

Wandering on the site, CPU goes back to normal after some time, but I couldn't find what the problem is with simple inspection (JS profiling says 99% idle).

Bug reproduced on Chrome.

Safari, which uses mp4 instead of webm, works fine :-)

Re:Introducing Paper by enneff on Thursday January 30, 2014 @09:23AM (#7150765)
The GP said "firefox/win7".
Re:Introducing Paper by weddpros on Thursday January 30, 2014 @10:06AM (#7150995)
I just checked: Firefox is also using the webm version of the video. It's just a strong hint, not a proof.
Re:Introducing Paper by teamonkey on Thursday January 30, 2014 @08:27AM (#7150522)
On a 4S you can literally see the battery meter run down before your eyes.
Re:Introducing Paper by blowski on Thursday January 30, 2014 @07:47AM (#7150326)
Doesn't work at all without JavaScript - just a black screen. And I don't have a Facebook account, so couldn't look at the mobile version.
Re:Introducing Paper by deckar01 on Thursday January 30, 2014 @10:21AM (#7151066)
The mobile version is unusable. The sideways scroll nav gets stuck between slides. The videos are corrupted or over compressed. The transitions lag and are so janky they only get about 3 frames rendered in time. The page doesn't even contain very much useful information about what Paper does.
Re:Introducing Paper by jrabone on Thursday January 30, 2014 @08:56AM (#7150651)
Then there's something wrong with either Firefox or your drivers; latest Chrome on a 4 year old machine (Windows 7 x64, Core i7, 12GB RAM) with a newer graphics card (GTX670) needs barely 10% CPU on a single core.
Re:Introducing Paper by erikj on Thursday January 30, 2014 @09:26AM (#7150774)
It managed to freeze Firefox on my laptop.
Re:Introducing Paper by danbee on Thursday January 30, 2014 @09:14AM (#7150724)
Yeah, I had this page open in Safari and it pegged a few cores.
Re:Introducing Paper by ape4 on Thursday January 30, 2014 @11:35AM (#7151546)
My laptop fan is still going after leaving the site.
Re:Introducing Paper by ulfw on Thursday January 30, 2014 @08:50AM (#7150623)
Wrong browser/OS then. On Safari/Mac it runs on an aging dual-core i7 Macbook Air with around 20% CPU
Re:Introducing Paper by brightghost on Thursday January 30, 2014 @09:35AM (#7150815)
...an "aging" ~two-year-old laptop with a top of the line processor?
Re:Introducing Paper by ulfw on Thursday January 30, 2014 @06:28PM (#7154124)
Well it's a 2011 Model. But calling a 1.8Ghz Dual Core 'top of the line' is a stretch, especially compared to the previous commenters 100% used octo-core.
Re:Introducing Paper by stusmall on Thursday January 30, 2014 @09:51AM (#7150908)
On Firefox/Ubuntu it pegged all cores on my Sandybridge i5 and the video never makes it past the envelope while the audio keeps going. Its hovering at about 40% CPU usage before the video even starts.
Re:Introducing Paper by acdha on Thursday January 30, 2014 @10:24AM (#7151091)
This is why WebM failed: on my (old) 2010 MacBook Air Firefox uses 100% CPU on two cores and it's still barely hitting 1fps with stuttering audio. Safari uses the native hardware-accelerated H.264 so it plays back perfectly at 15% of one core. Chrome's sandbox implementation appears to be less efficient so it's 80% but again, still perfect.

The vp8 developers clearly made some attempts at optimization – profilers show a lot of time being spent in an SSE3-optimized decode path – but that's a long, slow game trying to catch up with years of significant software and, later, hardware optimization.

Re:Introducing Paper by thejosh on Thursday January 30, 2014 @08:13AM (#7150446)
Same here with Chrome/arch.