Chrome Mostly Outperforms Safari 5 in JavaScript Benchmarks | Hacker News Reader

Chrome Mostly Outperforms Safari 5 in JavaScript Benchmarks

gavinroy.com

46 pointscrad25 comments

But: Safari 5 has hardware accelerated graphics on both Mac and Windows for HTML5 video. Which I imagine is as relevant to most people, even geeks like me, as a 10-20% speedup in Javascript.

For example, watch the video at http://videojs.com/ in full screen on Chrome and Safari. Last time I checked it looked way better in Safari. That said, Chrome team moves fast and maybe they will have it fixed soon.

If you're interested in progress you can track the issue here: http://code.google.com/p/chromium/issues/detail?id=38969

Not only video but also hardware-accelerated CSS animations and 3D transforms. Safari has the most advanced graphics architecture of any browser right now, and all the major browsers are working on catching up.

Firefox's plan: http://www.basschouten.com/blog1.php/2010/01/18/layers-cross...

IE's plan: http://blogs.msdn.com/b/ie/archive/2010/04/09/benefits-of-gp...

Opera hasn't publicly announced hardware-accelerated graphics yet, but they are moving to a new codebase that could support it in theory: http://my.opera.com/core/blog/2009/02/04/vega

Chrome actually seems like the least far along right now, which seems odd given their focus on performance. Some of the work was done for them by Apple in WebKit, but there's still a lot to be done.

According to this http://labs.opera.com/news/2009/12/22/ Opera already moved to the new codebase, but still no hardware acceleration at the moment.

> But: Safari 5 has hardware accelerated graphics on both Mac and Windows for HTML5 video.

IIRC Google Chrome use Google Skia to reader webpage and graphics, which is based on OpenGL.

I assume OpenGL is automatically hw accelerated?

Skia is not based on OpenGL and Chrome is not hardware accelerated. http://blog.chromium.org/2008/10/graphics-in-google-chrome.h...

I'm getting kind of sick about hearing about how browser X outperforms browser Y in Z JavaScript performance test. It's good to know that the browser developers are continuing to push JS performance--it's important that it gets better and better--but essentially every browser except maybe IE (haven't used it in years, can't speak to it) has comparable real-world JS performance.

I choose to use Safari because it's the browser that feels the most Mac-like, has ClickToFlash and 1Password, has great H.264 video performance, has a great interface, continues to support the newest web standards, never crashes, and is already familiar to me. As long as it isn't horrendously slower than other browsers, JS performance just isn't a concern to me--it has little real effect on my day to day browsing compared to everything I just listed.

The reason why I did this was because Apple is touting Safari 5 as being "The Worlds Fastest Browser" and specifically said "Powered by the Nitro JavaScript engine, Safari 5 on the Mac runs JavaScript up to 30 percent faster than Safari 4, 3 percent faster than Chrome 5.0,"

http://www.apple.com/safari/whats-new.html#performance

It more of a meta-comment about the genre of articles that yours loosely fit under :).

I missed the fact that you were actually referencing a claim made by Apple--sorry for jumping to conclusions!

It's important for resource-constrained devices. Think netbooks and smartphones.

Smartphones yes (though many smartphone users don't have much choice as to browser, or at least rendering engine), but netbooks are an arguable case.

I was referring more to people making their desktop browser choices based on benchmarks which have minimal, if any, effect on their day-to-day browsing. It seems like after every new browser release we get bombarded with "X desktop browser runs the SunSpider test faster than Y!" headlines, as if it's something that actually impacts people's browser choices.

... though WebKit's nightly outperforms both.

Chrome updates far more frequently than Safari, is it any surprise it's ahead slightly?

Considering they just released Safari 5 yesterday, it seems that it should be as up to date as recent Chrome releases.

Apple probably locked down on a stable build of WebKit weeks ago. Google has the luxury of staged rollouts with rollback if necessary, so they can take more risk with WebKit builds.

Is 'weeks' really that long? So we have to compare it to the version of Chrome released two days ago, and not last month?

I can see one problem with these benchmarks - you test dev version of Chrome and compare it with stable/released version of Safari...

Having said that - I'm using Chrome 6 on my Mac - it _feels_ faster then Safari 5.

Edit: updated to reflect 5.0.375 with little impact to the numbers.

I'm not sure that you should be using the "recommended tests" option of Dromaeo, since that option includes SunSpider and V8 Bench. We already know the results of those tests (since you ran them separately), so maybe you should just run the Dromaeo-specific tests.

Edit: Do you think it would skew things that much? It seems to show a much wider disparity with Chrome hitting > 5k runs per second and Safari 5 hitting only 308... My only concern about that is the appearance of cherry picking results instead of using their recommended tests.

One thing nobody ever mentions, but I always notice is how much more CPU power Chrome takes on my macbook. It's a few years old and anything remotely CPU intensive causes my fan to go nuts. I actually like Chrome better (I like the interface, and it's faster) but I stay with Safari because it doesn't hog my CPU...

Do we really care that much? I think people pick browsers for more reasons than just "speed".

My experience has been Safari 5 is more stable and snappier than Chrome on OS X. This outweighs the extra JavaScript execution speed in Chrome.

It's also much more stable on Windows for me. In Chrome a lot of websites just crash the tab (and bring down a few other tabs as well usually - so much about separate processes).

>so much about separate processes

I think the process is per-domain, not per-tab.

Personally, I use 4 different browsers at once when I'm on my Mac. Aside from the interfaces, I can't tell much of a difference between any of them. Sadly, I must admit that Firefox on the Mac is a bit lacking in polish.

It also doesn't crash when I use PivotalTracker.

Reply on news.ycombinator.com