Web Browser Grand Prix: Firefox 15, Safari 6, OS X Mountain Lion

[blppt]

Not that *I* would report you or anything (not being a hypocrite), but cant you guys get in trouble for blatantly advertising that you run Mountain Lion on a hackintosh system? Just wondering. I know Apple hasnt gone after the individual user yet, but when a fairly major site has a hackintosh system for testing, you would think you would get a call from their army of lawyers...

 

[blppt]

"To the people wondering about the performance on OS X, I think part of the performance difference is because Tom's is using a Hackintosh. My iMac, that doesn't have as good specs as the ones posted in the Test System specs, is faster across the board in the benchmarks I tried.

Apple knows the hardware it's going to use and so it optimizes for that hardware; on a Hackintosh you aren't going to see the same speed. If Tom's could do it, it would be interesting to see the stats when using an authentic Mac at as close to the same specs as possible."

Untrue. If it is set up *correctly* with the correct kexts being used, and somewhat more importantly, correct power management, there should be very little difference between a hack and a true mac. Go check out the geekbench or cinebench scores for hacks vs equivelent "true" macs.

That said, I would suggest toms run some other benchmarks on their hack box and make sure the cpu is scaling to turbo and normal states from idle. I know the sandy bridge i7s were having problems without a patched AICPM.kext where they will stay at idle speed (~1600mhz) rather than going to normal (3+ghz) and turbo under load.

 

[dooderoo]

[citation][nom]nitrium[/nom]So OSX is really just a LOT slower than Win7 generally for web browsing on the identical hardware. Is that right?[/citation]

Something is wrong with the Mac OS X numbers. Did a few of the tests on my much slower MB Air (2 Core 1.7MHz) and I get much better scores in each of the tests I tried. Sunspider on my machine is 190ms vs. 240ms here for example.

 

[Guest]

Baffled by Opera's catastrophic results. What's happened?

 

[kingius]

There is one HUGE thing missing from these tests - Canvas speed testing. I've been playing around with the three.js engine and noticed that when the render is using the canvas, Opera is by far the fastest browser. While Safari may achieve (for example) 6FPS when rendering out to the Canvas (and Firefox well below that), Opera is hitting 30FPS! So get this test in there because you're not showing a true picture of HTML 5 performance at all and secondly, it's the only way the other browser makers are going to improve their game.

 

[K2N hater]

Concerning Opera memory usage, it's all about design. Opera is known for having rather instant tab switching and that's because it does most of the rendering whenever the page loads, even if the tab is not selected. That's the reason it tends to take up much RAM. It's certainly not a flaw. Firefox, in comparison, takes a much different approach by doing a much partial rendering on all tabs but the active one, which greatly saves system resources (and that's particularly important on low-end PCs) but it comes with the price of making tab switching much slower.

 

[nebun]

[citation][nom]JOSHSKORN[/nom]How about 64-bit Internet Explorer 9 vs Waterfox 15.0?[/citation]
the web is not coded to take full advantage of 64bit browsers just yet....most sites will not even show the info correctly if you run a 64bit browser

 

[boingolover]

Regarding the hackintosh setup and the speed differences between windows and OSX, it looks like they're using an unsupported video card on their test machine, so it's possible that there is no hardware acceleration at all. Also, given that some browsers can use hardware acceleration for certain renderings and others can't, this could even potentially skew the results of the browsers on OSX, i.e., they might not be equally handicapped by a lack of hardware accel. A better test would either be using a card that's reported as well-supported on osx86project.org, or - better yet - running the tests on an actual apple computer as that is the only way you can run a vendor-supported configuration of both OSX and Windows, and you can easily dual boot using Boot Camp.