Google Chrome SunSpider benchmark results

I promised to post the SunSpider benchmarks for Google Chrome, and boy, V8 surely blew away other browsers in many aspects! For the visual learners, here's a pretty chart to show just how monstrous Chrome is when it comes to JS perfs (as Intchanter noted, smaller is better):

Here are some nicely formatted raw numbers:

  Chrome Firefox 3.0.1 Safari 3.1 WebKit nightly Opera 9.52 IE 7
3d 110.2 357.2 412.8 371.8 418.6 1287.6
access 83.8 494.8 531.4 346.8 612.2 1887.4
bitops 61.2 336.8 421.8 187.6 550.2 1718.4
controlflow 3.6 40 93.8 18.8 53.4 484.6
crypto 54.2 200.4 250.2 128.2 227.8 1022
date 265.2 254.2 343.8 197 631.4 874.6
math 110.2 357 450.2 331.6 337 1177.8
regexp 370.2 225.6 303.2 265.6 465.4 375
string 511.2 719 721.6 487.8 985.2 18296.8
If you'd like to hack on the V8 JavaScript engine or play around with the V8 shell and you're on a Mac, I encourage you to take a look at this post.

Just for clarification,

Just for clarification, these numbers are in milliseconds, so smaller is better.

This is as opposed to some other benchmarks which report N operations per second, where larger is better.

Was squirellfish used in

Was squirellfish used in webkit's nightly build?

Yes.

Yes.

A chart of your data with

A chart of your data with the IE outlier excluded: chart

Also, the IE strings data seems a bit strange. In your chart it seems an order of magnitude faster than it does in the data. Perhaps I'm reading it wrong.

Oh yes, the order of

Oh yes, the order of magnitude was so huge I simply brought IE's string ms value down to 1900 to keep the proportions right on chart. So you're not reading anything wrong.

Would it be a lot of effort

Would it be a lot of effort to include Konqueror/KHTML (in its 4.1 version with overhauled bytecode-compiling KJS engine) in the benchmarks?

It would be nice if you

It would be nice if you included results for the new TraceMonkey JavaScript VM that is present in the latest Firefox nightlies. It should be quite comparable to the V8 VM found in Chrome.

Just grab a nightly from http://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/latest-tracemonkey/ , enable the VM in about:config (search for "jit"), restart Firefox and do the benchmark.

firefox developers has made

firefox developers has made a benchmark for it.

http://weblogs.mozillazine.org/roadmap/archives/2008/09/tracemonkey_upda...

I did try to do that. But as

I did try to do that. But as Brendan said...

Tonight's builds include a fix for the bug that ilooped a SunSpider test (my apologies to those of you who suffered that bug's bite).

I was one of those people who suffered that bug's bite.

Wow, I had no idea IE 7's JS

Wow, I had no idea IE 7's JS performance was so poor. Yet another reason that browser is garbage.

i'm willing to try it out

i'm willing to try it out just to see if it works more efficiently than FireFox... if it's faster than Firefox and isn't IE, then i'll use it

Chrome is not always faster

Chrome is not always faster than Firefox.

There's an example here where Firefox beats out Chrome significantly in speed and memory usage:
http://www.wrensoft.com/forum/showthread.php?t=2789

We could do with more real life benchmarks. SunSpider, as pointed out, is developed by the WebKit team - the people behind the engine for Chrome. Not to say it is completely biased, but it is likely that the Chrome team would have tested and optimized their code against SunSpider to make sure it runs well. How well this reflects in real life applications is reallly more important.

Chrome uses a completely

Chrome uses a completely different JavaScript engine than Webkit so if anything SunSpider should be biased *against* V8 (chrome's JavaScript engine).

Sunspider is one of the few credible Javascript benchmarks out there, so it's likely that *every* browser is developed taking into account Sunspider's results - there's certainly no reason why anyone developing a Javascript engine should ignore it.

That said; Safari, Opera and Firefox all have preview builds available with dramatically improved JavaScript performance available which are at least as reliable as Chrome at the moment.

My tests in XP between the current crop of prereleases for Opera, Safari (squirrelfish), Firefox (tracemonkey), IE and Chrome suggests that Chrome handily spanks all comers in JavaScript speed, but has problems with a lot of pages and some plugins that work fine with all the others, and despite the apparent speed advantages doesn't seem any faster when working with demanding real world content like the SproutCore "Photos" demo.

Thank you very much for such

Thank you very much for such an interesting article

Having nothing better to do

Having nothing better to do tonight, I decided to run my own series of tests using sunspider to determine which browser was the fastest on MY computer -- a Dell Q6600 with 6 gig of Ram -- using Vista 64.

Firefox 3.1b1pre : 1301 ms
Chrome .2.149.29 : 1595 ms
IE 7 : 22,513 ms
Opera 9.52 : 3979 ms
Safari 3.1.2 : 3547 ms

I was rather suprised at IE's time. I knew it was going to be the slowest but not by that big of a margin.

Were you using the 32-bit

Were you using the 32-bit (normal) mode of IE when testing?

All browsers were 32bit.

All browsers were 32bit.

Wilson, Could you post the

Wilson,

Could you post the basic specs of the system you ran the tests on (processor and amount of RAM)?

Absolutely. This was ran on

Absolutely. This was ran on a 2.2 GHz Intel Core 2 Duo machine with 1.5 GB of RAM.

Absolutely rocks in every

Absolutely rocks in every way. Google has once again proven that it has what is needed to deliver great products. it’s a decent browser…download time may be a dash slower than other browsers but the app load time, smooth rendering, space utilization etc are great.

I added firefox 3.1 to these

I added firefox 3.1 to these stats and the resulting charts + sheets can be found on http://spreadsheets.google.com/pub?key=pmYpIvbY8QX-8WEutszwwHA

PS. To the author of this site, seriously add a better spam filter

@David: I'm currently using

@David:
I'm currently using Mollom as my spam filter, and, trust me, it works way better than Akismet. I literally had dozens of spam to moderate every week when I used Akismet, and now, after using Mollom, only have around half a dozen over the span of several weeks. It still misses some blatantly obvious spam (in Arabic), perhaps because it's a relatively new kind.

Another problem I'm seeing recently is that people are getting paid to post what I call "quality spam." Their responses are completely on topic and even insightful at times, but they fill in their homepage with sites that are paying them to get more ranking on Google. The Computer Chess guy / gal right above is a live example of "quality spam." I usually just remove the link to their so-called "homepage". I should probably set up rel="nofollow" some time and warn readers about it. This is the kind of spam that Mollom has a hard time to deal with.

On the topic of your benchmarks, I find it fun to see that you did what I did to the IE results for strings. IE was so slow for strings that I simply trimmed down the number to prevent the chart from being highly disproportional.

A month later and Chrome

A month later and Chrome usage is way down. People tried it, liked it (generally), then went back to their usual browser. Chrome IS a good browser, but it's too bare bones at the moment - when you see all the Firefox plugins available (arguably THE browser that is competing with Chrome - IE users just don't switch no matter what!).

I'd prefer you to call them

I'd prefer you to call them Firefox "extensions" instead of "plug-ins". For Firefox, those are two different stuff.

I am an IE user who do

I am an IE user who do switch to Firefox! So I dont know what you are trying to say. I only use IE to browse porn now! nothing else!

lol.. IE to watch p0rn???

lol.. IE to watch p0rn??? you'd be better off with Safari or Chrome with private browsing...

Ditto.

Ditto.

Wow, I didn't realize that

Wow, I didn't realize that IE was that ridiculously slow. I'm also pretty surprised that Opera performed so poorly, guess I'll give up on Opera...again.
But still, IE being behind by that large of a margin is absurd. I knew that IE was slow, but not that slow. Frankly, I would think that Microsoft would see this issue and at least make an attempt to boost IE's speed, but guess not, Microsoft is going downhill.

Yeah...about IE to browse porn, maybe you should bring it up somewhere else? Just hinting...

For the record, many of the

For the record, many of the sunspider benchmarks are incorrectly coded, and assumes invalid things that cause IE to process the commands (correctly btw) differently than other browsers. This is especially apparent in the string tests which IE is correctly returning errors. In another example, the base64 string tests are failing under multiple browsers in different ways, just IE while it fails takes longer to finish executing.

This benchmark is quite honestly poorly written, unreliable, and obviously hasn't ever been tested (or perhaps it was, but the creator chose to ignore it since it makes IE look bad). Go ahead, check out the source to the benchmark yourself, it's pretty obvious when you actually look at it.

Google Chrome isn't

Google Chrome isn't available for Linux yet. I'm running Ubuntu 8.04 on both of my laptops. Kind of surprising that Google is so gung-ho about open source software, but didn't release a version that runs on an open source OS. Has anyone seen a release date for a Linux version? Firefox 3 is quite fast and stable for me in Ubuntu. Plus, there are all the awesome addons. I don't see any other browser replacing Firefox as my primary browser.

Would it be a lot of effort

Would it be a lot of effort to include Konqueror/KHTML (in its 4.1 version with overhauled bytecode-compiling KJS engine) in the benchmarks?

I tried running the test in

I tried running the test in Konqueror but it never got to the results page. I'm pretty sure it silently crashed since it was going through the tests at a decent speed (read: faster than IE) but around the end the page just went blank.

I am an IE user who do switch

I am an IE user who do switch to Firefox! So I dont know what you are trying to say

A large percentage of IE

A large percentage of IE users just don't want to switch. I'm guessing that it's because 1) Windows is the most common desktop OS by a very large margin, 2) IE comes bundled with it, and 3) because of 1 and 2, average people who just use computers for simple/common purposes don't care to customize their computers to a large extent, and since IE works (it shows you web pages and it will run almost anything, albeit slowly) they don't care to look for something better.

interesting article ,

interesting article , thanks.




Post new comment

The content of this field is kept private and will not be shown publicly.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <code> <ul> <ol> <li> <dl> <dt> <dd>
  • Lines and paragraphs break automatically.

More information about formatting options

CAPTCHA
This question is for testing whether you are a human visitor and to prevent automated spam submissions.