• Are Mac Browsers Really Slower?

    February 19th, 2005

    At the heart of the claim that Macs are slower than Windows PCs is the perception that browsers run slower. A lot of that, of course, is based on the fact that Internet Explorer generally works a lot faster under Windows, simply because it’s optimized for that platform. Yet when I’ve done benchmarks of this sort in the past on powerful hardware for both platforms, the results have been largely a wash. If there’s a difference, it’s not significant enough to fret over. But my tests were quite basic, and I was more interested in how that hardware handled hard-core Photoshop rendering functions.

    Of course, the arrival of Apple’s Safari enters into the equation, since it is one of the browsers that’s supposedly faster than the rest of the competition. But now it appears that Opera 8.0, available in preview form on the Mac, may have finally earned that crown.

    Or at least that’s what two recent articles on the subject claim.

    As with some other sets of benchmarks, there are terms and conditions that may make the results a little questionable. In the latest example, Dan Knight, of LowEndMac fame, assembled a collection of browsers from Linux, Mac OS 9, Mac OS X, a prerelease version of Tiger, and Windows XP, and did a series of six sets of timed runs on each. It clear that Dan did an exemplary job of selecting applications and test suites. The real shortcoming is that none of the tests were done on anything resembling current Mac or Windows hardware, although it doesn’t necessarily mean the results won’t scale up. But I digress.

    If you take the conclusions at face value, you’ll come away a little disappointed. For the most part, Windows browsers handled most tests faster. Mac OS X, however, excelled at images and the display of browsing history. You’ll also be surprised that the highly touted Internet Explorer killer, Firefox, isn’t really faster than Internet Explorer under Windows, although it’s more secure and renders some sites more accurately. Overall, however, Opera, despite being in pretty rough shape, wins the speed crown, except in launch and relaunch times. I’m very curious to see how it fares once the development process is further along.

    Another set of kudos for Opera comes from Charles Moore in one of his endless Mac OS X Odyssey columns.

    At the same time, Safari acquitted itself well, although I won’t pay any attention to the tests run on a prerelease version of Tiger. It’s not reasonable to expect that it has been optimized for performance. And I won’t comment on how Apple might feel about such things. You and I both know the answer.

    There are other interesting tidbits in Dan’s tests, such as the fact that, in general, application relaunch times tend to be a lot faster under Windows, followed by Linux. But this may be a reflection of the speed of the hard drive in retrieving cached data. And here’s where the tests may be flawed. You see the test hardware consisted of a 400MHz Power Mac G4 and an 800MHz Pentium III. Both were outfitted with 256MB of RAM, and you know that’s a pretty basic environment for Mac OS X. In fact, many claim it’s just not acceptable. In addition, it’s not at all clear if any effort was made to optimize performance, such as repairing disk permissions and cleaning cache files on the Mac. And Dan never tells us which drives were used.

    It’s not that Dan doesn’t realize the nature of the shortcomings. He writes: “It would be nice to see comprehensive benchmarks using more modern hardware—single- and dual-processor computers, G5 and Pentium 4 CPUs, different hardware configurations (512MB vs. 256MB RAM, faster vs. slower hard drives, different video cards, etc.), and Linux on a Mac vs. Linux on Intel hardware.”

    So where does that leave us? Well, if the computers were matched up a little better, with identical drives and 512MB of RAM, I might take them a little more seriously, at least in the cross-platform comparisons. But in the end, I think Dan owes us all an attempt to acquire systems that represent what you can actually purchase in 2005. No, not necessarily workstation class computers, but perhaps something akin to what an average business user might buy.

    In short, I don’t necessarily think that Windows has been shown to be a better Internet platform than the Mac. It’s true that Dan clearly has taken a lot of care was in collecting industry standard tests and lining up as complete a lineup of browsers as I’ve ever seen. But I can’t get past the hardware limits, except for comparisons on the same platform of course.

    Having gone through all this work, it would be nice to see Dan buy or borrow a proper set of recent hardware and rerun his tests. On the other hand, he’s dropped the gauntlet. And if he isn’t able to do a follow-up, it’ll be up to others with an equal set of patience to take these tests to the next level. And, no, don’t ask me right now. As I said, it requires someone with a lot of patience.

    | Print This Article Print This Article

    Leave Your Comment