backtop


Print 39 comment(s) - last by Smilin.. on Sep 14 at 12:55 PM

The conclusion of our three part series looking to expose each next gen browser's greatest strengths and weaknesses

In our exploration of next generation browsers we first examined the user interface, installation details, time to install, and application launch times.  We next turned our attention to CPU and memory usages, as well as comparing and contrasting the security of the various browser offerings.  Topics we haven't looked at yet include rendering comparisons, synthetic benchmarks, plug-ins, and standards support, as well as our conclusions.  We planned just one more piece, but that's a lot of ground to cover, so we've broken it up into two more pieces. 

In this first piece, we'll look at synthetic benchmarks and plug-in support.  In the next (and final) segment, we'll examine rendering performance, standards support (including performance in the Acid3 benchmark), and our conclusions on the state of the browser war and who we believe the current winner(s) is/are.

6.  Synthetic Benchmarks:


The first synthetic benchmark we ran was Celtic Kane's Javascript test.  The homepage for the test showed Safari 4 to be the leading web browser.  Our own testing indicated that it has been passed by Google's Chrome.  Safari came in second, Opera in third, Firefox in fourth, and Internet Explorer came in at a distant fifth.


The next synthetic benchmark run was the popular Sunspider Javascript test.  This test again showed Chrome 4 beating Safari 4.  This time, though, Firefox was a runner up.  Opera performed unexpectedly poorly in this test, though Opera 10 managed a bit better performance.  And Internet Explorer 8 performed the worst of all, taking nearly eight times as long as Chrome 4 to complete this test.  Combined with Celtic Kane, these tests indicate the Safari's Squirrelfish engine, Chrome's V8 to be the clear JS leaders and the JS performance of Trident, IE 8's engine, to be dismal. 


Our next synthetic benchmark was Peacekeeper, a Futuremark test suite which we recently profiled.  Our testing with the suite indicated Chrome to be in the lead, followed closely by Safari and then Firefox.  Opera 10 managed middling results.  Meanwhile, Internet Explorer was the slowest, managing a mere sixth of the score of Chrome.  The suite looks at a number of aspects, including JS performance, CSS performance, rendering, and more, so it is a good general indicator of speed.


The last synthetic benchmark we used was How-To-Create UK's CSS test.  The test loads approximately 2,500 DIVs and times how fast the load takes.  Unfortunately, WebKit browsers (Safari and Chrome) aren't supported due to how they measure time, so we could only get results for Firefox, IE 8, and Opera.  For these browsers,  Opera 10 barely led Firefox 3.6a1, while Internet Explorer 8 yet again lagged in performance.

Conclusions to be drawn from the synthtetic tests -- Chrome is the fastest browser with Safari close behind.  Opera and Firefox are just slightly behind.  And Internet Explorer 8 is the slowest browser.  Even on content heavy sites, though, this performance difference is not as great as these tests might seem to indicate.  In fact, it may only account for a couple of extra seconds of load time.  Still, it could become an annoyance on content-heavy pages like Facebook.

7.  Plug-ins

A common misconception is that Firefox is the only browser that has plug-ins, add-ons, extensions, or otherwise named optional components.  Plug-ins/add-ons/extensions are, in fact, a vital part of modern browsers.  There are a diverse variety of formats include ActiveX, NPAPI, Java, Google Gears, RSS, and Atom. 

Mozilla Firefox does arguably lead in this field, having the most enthusiastic developer community for plug-ins, and the most useful plug-ins.  Firefox 3.6a1 does not provide is support for classic Java plugins and ActiveX plug-ins. 

Looking at the other browsers, Opera (which does not support Gears, ActiveX), Chrome (also no ActiveX), and Safari (no ActiveX, Gears partial only) all offer decent plug-in support, but their developer communities are still in a fledgling state.  Internet Explorer 8, on the other hand, offers a lot of plug-ins -- in fact, plug-ins are essential to improving the browser's standards support.  Unfortunately IE 8 does not support the NPAPI extension language commonly used for Mozilla.  IE 8 plug-ins are generally more for utility rather than aesthetics.

Plug-ins are most accessible to Firefox users, as Firefox's plug-in system is friendly enough for even beginning users.  Nonetheless, if other factors make you pick another browser, it's a good idea to check out what kind of plug-ins are available for it, as there will surely be some useful ones.

Note: All benchmarks were performed in 32-bit Vista on a Sony VAIO laptop with 3 GB of RAM, a T8100 Intel Processor (2.1 GHz), and a NVIDIA 8400 GT mobile graphics chip. The number of processes was kept consistent and at a minimum to reflect stock performance.


Comments     Threshold


This article is over a month old, voting and posting comments is disabled

RE: Ignoring
By Roffles on 9/10/2009 10:38:28 AM , Rating: 2
The biggest flaw of all is the notebook he's using. Understandably, I get entirely different numbers for some of these benches on my Q9550 desktop system. But the differences aren't as extreme....telling me these numbers are exposing the hardware limitations of his computer rather than the software limitations of the browsers.


RE: Ignoring
By foolsgambit11 on 9/10/2009 12:55:56 PM , Rating: 2
While I agree that there is little value in single-platform benchmarks, I disagree with your analysis of your benchmarks versus his. Unlike in hardware benchmarking, where you don't want any other hardware to bottleneck the product being tested, in software benchmarking it can be very valuable to see how the software handles resource starved situations. I'm not saying that Jason's single benchmarks on an average laptop accomplish that, by any means.

I don't think that these numbers are exposing the hardware limitations of his computer. They are exposing how each browser responds to his specific hardware limitations. In some ways, that's more valuable for software benchmarking than testing on a high-performance rig, since it gives the average Joe an idea of the performance he can expect.


RE: Ignoring
By Targon on 9/10/2009 11:19:28 PM , Rating: 2
The problem here is that speed(or lack of it) can come from multiple sources, so looking a bit deeper at WHERE the delays come from is important.

If the browser tends to use a lot of hard drive space for temporary files, or is slow to check for and load cached images, the speed of the hard drive will cause reduced performance. This is why the memory footprint needs to be looked at along side performance vs system memory to see WHERE the real speed of the browser comes from. A browser that is designed to keep as little in RAM as possible MIGHT use the hard drive more then.


RE: Ignoring
By gstrickler on 9/10/2009 1:16:27 PM , Rating: 2
The differences in your results aren't as extreme in absolute terms or in percentage terms? A faster machine will have much small absolute differences, but the percentages should be similar. On a "fast enough" machine, the differences would be below the level of perception, but that doesn't mean there aren't important difference between these browsers.

Most users don't have quad core machines, in fact, since laptop/notebook machines now outsell desktops, the machine Jason used is a much better representation of what a typical user will have.


"Game reviewers fought each other to write the most glowing coverage possible for the powerhouse Sony, MS systems. Reviewers flipped coins to see who would review the Nintendo Wii. The losers got stuck with the job." -- Andy Marken














botimage
Copyright 2014 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki