Substantial Javascript improvements with iOS 6 [Archive]

PDA

View Full Version : Substantial Javascript improvements with iOS 6

MythicFrost

Jun 12, 2012, 11:21 PM

PurrBall

Jun 13, 2012, 12:28 AM

I've noticed the Facebook app running much smoother, has this new JS engine been enabled for apps?

MythicFrost

Jun 13, 2012, 12:51 AM

I've noticed the Facebook app running much smoother, has this new JS engine been enabled for apps?
No no, apps don't use Javascript. It's just for web browsing.

PurrBall

Jun 13, 2012, 01:27 PM

No no, apps don't use Javascript. It's just for web browsing.

I'm almost positive Facebook uses a WebView, and, in turn, JavaScript.

Menel

Jun 13, 2012, 02:01 PM

MythicFrost

Jun 13, 2012, 08:47 PM

I'm almost positive Facebook uses a WebView, and, in turn, JavaScript.
Nah, it's definitely not a UIWebView. That aside, third party apps didn't receive the Nitro 2 JS engine for Safari in iOS 4.3, so they probably won't have received this either.
Mine is slow =/ oh well, serviceable for now.
What's your 4 scoring?

Anti-Lucifer

Jun 13, 2012, 09:20 PM

On SunSpider 0.9.1 my iPad 3 got 1426ms and my 4S got 1716ms, and on RightWare BrowserMark my iPad 3 got 128624 and my 4S got 107914.

Here's Anandtech's preview (http://www.anandtech.com/show/5810/samsung-galaxy-s-iii-performance-preview) of the SGS3 which shows benchmarks of SunSpider and BrowserMark for the SGS3 and other devices, including the 4S on iOS 5.

Apple software optimisations > anything else on the market, bar none. My iPad 3 with dual core cpu + ios6b1 can render sun spider 1424ms as well. Now you tell me which is more efficient?

Sunspider iOS6b1 iPad 3 results (http://www.webkit.org/perf/sunspider-0.9.1/sunspider-0.9.1/results.html?%7B%22v%22:%20%22sunspider-0.9.1%22,%20%223d-cube%22:%5B66,70,74,65,68,82,64,63,77,71%5D,%223d-morph%22:%5B53,53,53,53,53,52,53,52,53,53%5D,%223d-raytrace%22:%5B94,99,97,97,94,95,94,95,97,94%5D,%22access-binary-trees%22:%5B17,17,20,21,18,21,18,18,21,17%5D,%22access-fannkuch%22:%5B54,53,54,54,53,53,53,54,58,54%5D,%22access-nbody%22:%5B27,27,27,26,27,26,26,26,26,27%5D,%22access-nsieve%22:%5B28,28,29,29,28,29,28,29,29,29%5D,%22bitops-3bit-bits-in-byte%22:%5B5,4,5,5,5,5,4,5,5,5%5D,%22bitops-bits-in-byte%22:%5B18,18,18,18,17,17,17,17,18,17%5D,%22bitops-bitwise-and%22:%5B32,33,33,34,33,33,32,32,33,32%5D,%22bitops-nsieve-bits%22:%5B21,21,20,21,21,20,21,21,21,21%5D,%22controlflow-recursive%22:%5B17,17,17,16,16,16,17,17,16,16%5D,%22crypto-aes%22:%5B85,74,74,74,74,74,86,85,73,87%5D,%22crypto-md5%22:%5B32,32,31,39,32,31,31,32,33,31%5D,%22crypto-sha1%22:%5B23,23,22,22,35,22,22,29,23,22%5D,%22date-format-tofte%22:%5B110,118,105,105,106,104,108,107,103,107%5D,%22date-format-xparb%22:%5B107,109,120,106,109,121,109,109,123,108%5D,%22math-cordic%22:%5B26,26,26,26,26,26,25,26,26,26%5D,%22math-partial-sums%22:%5B59,60,60,60,60,60,59,60,59,59%5D,%22math-spectral-norm%22:%5B21,20,22,21,22,21,21,21,21,21%5D,%22regexp-dna%22:%5B76,77,76,89,75,76,76,75,75,76%5D,%22string-base64%22:%5B44,44,44,48,43,43,44,44,44,52%5D,%22string-fasta%22:%5B69,68,68,69,68,72,68,69,68,69%5D,%22string-tagcloud%22:%5B87,100,87,89,100,87,87,87,87,90%5D,%22string-unpack-code%22:%5B158,161,157,156,161,156,160,158,157,159%5D,%22string-validate-input%22:%5B77,86,92,78,88,89,90,77,90,79%5D%7D)

MythicFrost

Jun 13, 2012, 09:39 PM

Apple software optimisations > anything else on the market, bar none. My iPad 3 with dual core cpu + ios6b1 can render sun spider 1424ms as well. Now you tell me which is more efficient?

Sunspider iOS6b1 iPad 3 results (http://www.webkit.org/perf/sunspider-0.9.1/sunspider-0.9.1/results.html?%7B%22v%22:%20%22sunspider-0.9.1%22,%20%223d-cube%22:%5B66,70,74,65,68,82,64,63,77,71%5D,%223d-morph%22:%5B53,53,53,53,53,52,53,52,53,53%5D,%223d-raytrace%22:%5B94,99,97,97,94,95,94,95,97,94%5D,%22access-binary-trees%22:%5B17,17,20,21,18,21,18,18,21,17%5D,%22access-fannkuch%22:%5B54,53,54,54,53,53,53,54,58,54%5D,%22access-nbody%22:%5B27,27,27,26,27,26,26,26,26,27%5D,%22access-nsieve%22:%5B28,28,29,29,28,29,28,29,29,29%5D,%22bitops-3bit-bits-in-byte%22:%5B5,4,5,5,5,5,4,5,5,5%5D,%22bitops-bits-in-byte%22:%5B18,18,18,18,17,17,17,17,18,17%5D,%22bitops-bitwise-and%22:%5B32,33,33,34,33,33,32,32,33,32%5D,%22bitops-nsieve-bits%22:%5B21,21,20,21,21,20,21,21,21,21%5D,%22controlflow-recursive%22:%5B17,17,17,16,16,16,17,17,16,16%5D,%22crypto-aes%22:%5B85,74,74,74,74,74,86,85,73,87%5D,%22crypto-md5%22:%5B32,32,31,39,32,31,31,32,33,31%5D,%22crypto-sha1%22:%5B23,23,22,22,35,22,22,29,23,22%5D,%22date-format-tofte%22:%5B110,118,105,105,106,104,108,107,103,107%5D,%22date-format-xparb%22:%5B107,109,120,106,109,121,109,109,123,108%5D,%22math-cordic%22:%5B26,26,26,26,26,26,25,26,26,26%5D,%22math-partial-sums%22:%5B59,60,60,60,60,60,59,60,59,59%5D,%22math-spectral-norm%22:%5B21,20,22,21,22,21,21,21,21,21%5D,%22regexp-dna%22:%5B76,77,76,89,75,76,76,75,75,76%5D,%22string-base64%22:%5B44,44,44,48,43,43,44,44,44,52%5D,%22string-fasta%22:%5B69,68,68,69,68,72,68,69,68,69%5D,%22string-tagcloud%22:%5B87,100,87,89,100,87,87,87,87,90%5D,%22string-unpack-code%22:%5B158,161,157,156,161,156,160,158,157,159%5D,%22string-validate-input%22:%5B77,86,92,78,88,89,90,77,90,79%5D%7D)
They're pretty good optimisations but we may see Google's comeback with uh... Jellybean? They tend to trade blows which is a good thing.

Daveoc64

Jun 14, 2012, 07:00 AM

Nah, it's definitely not a UIWebView.

It is.

The Facebook app is the same on both iOS and Android.

The "App" simply wraps the mobile version of the site with things like support for push notifications.

MythicFrost

Jun 14, 2012, 07:24 AM

It is.

The Facebook app is the same on both iOS and Android.

The "App" simply wraps the mobile version of the site with things like support for push notifications.
Heh, it seems you (and PurrBall) were right. I've not looked at the mobile version of the site in a while, but yes it's exactly like in the app -- though I think some of the buttons in the app are actually buttons and not a part of the WebView.

Though I still don't think these JS improvements have been passed on to third party web browsers. I ran the same SunSpider test in SkyFire and I got 6844 ms.