WINDOWS AND OS X EACH HAVE THEIR SUPPORTERS, BUT WHICH WILL MAKE YOUR APPLICATIONS RUN AT FULL SPEED? DARIEN GRAHAM-SMITH FINDS OUT
Apple hardware is everywhere, and if you don’t already own a Mac, you may well be considering one for your next PC. In our reviews, the MacBook Pro 13in with Retina display (see p43), 27in iMac and MacBook Air received Recommended awards. Even in business, Mac desktops have become a viable choice (apart from their higher pricetag compared to Windows competitors), thanks partly to their ability to run both OS X and Windows, whether via a virtualisation package such as Parallels, or via Apple’s Boot Camp dual-boot system.
But which system should you make your main operating system? OS X has the advantage of better security and better integration for things such as multitouch gestures and function keys, but Windows has its own strengths, including more games, wider hardware compatibility and broader support for legacy software One factor that’s difficult to quantify is performance. OS X and Windows are based on different kernels, with distinct approaches to the likes of multitasking and virtual memory.
What’s more, while many mainstream applications are offered on both platforms, they’ll necessarily be implemented in different ways, as dictated by the different platform architectures. As such, they perform the same jobs quite differently. With that in mind, we set out with a stopwatch to time how long OS X and Windows took to complete a variety of common desktop tasks. Our mission was to find out whether Apple’s native OS gives a performance advantage over Windows, or if it’s actually slower.
To ensure our results were representative, we tested the OSes on the same hardware – a pair of midrange Mac systems with relatively limited power, where performance could easily be a real-world issue. One was a 2008 iMac with a 2.4GHz Core 2 Duo E8135 processor, 3GB of 667MHz DDR2 RAM and a 250GB Hitachi Deskstar P7K500 3.5in hard disk. The other was a 2011 MacBook Air, with a 1.6GHz Core i5-2467M processor, 4MB of 1333MHz DDR3 RAM and an Apple SM128C SSD.
Both machines were set up as Boot Camp dual-boot systems. Since most Mac users keep their operating system current, we used the latest release of OS X 10.9, Mavericks; for Windows, we used the most popular version of the OS, namely Windows 7 Home Premium, running natively on the hardware.
Test 1: Browser performance
 |
| GRAPHS 1 AND 2: BUNDLED BROWSER PERFORMANCE |
We started our tests by looking at web-browser performance. This is an important aspect of overall performance, since these days we use our browsers for everything from sending and receiving email and working on documents to watching movies and playing games. To get an all-round picture, we tested each platform with five benchmarks.
The SunSpider, Kraken and Octane tests focus on JavaScript performance, reflecting the general responsiveness of apps such as Gmail and Google Drive. The CanvasMark and Peacekeeper benchmarks assign more weight to the graphical and entertainment capabilities of HTML5, giving an indication of each platform’s multimedia performance. We first carried out these tests using the browser bundled with each OS – Safari 7 on OS X and Internet Explorer 10 on Windows 7.
You can see our results in the graphs above, which detail the tests across both browsers on both platforms. Graph 1 shows scores from the Octane, CanvasMark and Peacekeeper benchmarks: these all return absolute scores, so taller bars show better performance. The Kraken and SunSpider benchmarks – shown in graph 2 on the right – return results in milliseconds, so here a lower score is better.
Overall, we found Safari on OS X took the lead. On our iMac, across all benchmarks, Safari proved 67% faster on average than Internet Explorer on Windows. On the MacBook Air, Safari was around twice as fast overall.
We saw one glaring anomaly: Microsoft’s browser pulled far ahead of Safari in the SunSpider test. We suggest you don’t attach too much significance to this, however: when you consider IE’s laggardly scores in the other benchmarks, it looks as though the browser has been specifically optimised for this benchmark. We expect real-world performance will be closer to what we’ve seen in the Kraken and Octane tests.
Clearly, if you’re planning to stick to the OS developer’s own browser, Safari on OS X is the way to go, but it’s worth remembering that these native browsers aren’t your only options. One popular alternative is Google Chrome, which is offered for both OS X and Windows.
 |
| The Peacekeeper benchmark tests the gaming capabilities of web browsers |
When we repeated our tests using the latest build of Google Chrome (version 29) on both systems, we saw OS X’s apparent advantage completely erased. In fact, Apple’s platform was left slightly behind overall – the iMac trailed across our tests by an average of 8%, the MacBook Air by around 1%. This is probably a small enough margin to live with, but the upshot is clear: when it comes to online performance, your choice of OS makes much less difference than your choice of browser.
Test 2: Office applications
For our next set of tests, we turned to two of the most widely used applications in the world – Microsoft Word and Microsoft Excel. Microsoft publishes OS X editions of these programs as part of the Office for Mac suite, but they have different interfaces. Also, while Windows users are on Office 2013, the latest Mac suite is the 2011 edition.
 |
| GRAPH 3: NATIVE WORD-PROCESSING SOFTWARE PERFORMANCE |
This being the case, we expected to see some palpable differences in the user experience across the two platforms. Basic tasks such as typing, printing, opening files and calculating sums happened so smoothly in both operating systems that we couldn’t detect any difference. But larger tasks, such as big find-and-replace jobs and drawing graphs, make differing levels of performance noticeable.
We started by opening a large document in Word – the full text of Lewis Carroll’s Alice’s Adventures in Wonderland, to be precise – and timed how long it took the software to replace every letter “I” with the word “TEST”. This isn’t exactly a real-world exercise, but it’s an indicator of overall performance.
Then, we pumped up the font size to 32 points and timed how long it took for Word to repaginate the document. In Word 2013 on Windows, this stretched it from 43 pages to 587 pages, while in Word for Mac 2011, owing to differences in the default formatting, the document ran to 683 pages. Our results, in seconds, can be seen in graph 4, below.
Clearly, Word for OS X is slower than Word for Windows. On the MacBook Air, the difference wasn’t too alarming – as a proportion, eight seconds is much longer than five seconds for a find-and-replace operation, but our test used a deliberately extreme usage case. In real-world use, the difference is likely to be negligible. On the iMac, however, Word 2011 for Mac lagged by a greater margin, especially in the repagination task.
As with our browser test, Word isn’t your only choice. If you’re using a Mac for work, you might prefer to use a Microsoft-branded office suite. But for many Mac users, Apple’s own iWork suite – comprising Pages, Numbers and Keynote – provides sufficient power. It’s cheaper, too, working out at $63 for the three applications, versus $170 for Microsoft’s Office for Mac Home & Student edition. A more appropriate comparison might be Word 2013 versus Pages 9.3. The results of that test are shown in graph 3, above. As you can see, the situation is flipped: Microsoft Word is quicker on Windows than on a Mac, but Pages on OS X delivers the best performance on any platform.
We also compared performance in Excel. Again, we started by testing the latest versions, pitting Excel 2013 on Windows against Excel 2011 on OS X. This time, the test was to render a stacked 3D bar graph representing four series of 1000 random numbers. Here, we saw no effective difference between versions, so we moved on to compare Excel on Windows with Apple’s Numbers on OS X. Since its charting options don’t precisely match those of Excel, our approach doesn’t allow a perfect comparison of performance.
However, we can compare how long it takes Excel and Pages to render stacked bar graphs in their respective default 2D and 3D forms, which gives us an idea of how long a typical user will spend waiting for their graph to draw or update. The results – shown on graph 5, below – are the opposite of what we saw with Pages.
 |
| GRAPH 4: MICROSOFT WORD PERFORMANCE |
 |
| GRAPH 5: NATIVE SPREADSHEET SOFTWARE PERFORMANCE |
For journalists and authors (and other word-heavy professions), the blazing responsiveness of Pages makes OS X and iWork an attractive pairing, with Microsoft Office on Windows taking second place. If you work heavily with figures, the situation is reversed, thanks to Numbers’ slower handling of graphs.
Test 3: Adobe Acrobat and Photoshop
 |
| GRAPH 6: ACROBAT XI PRO PERFORMANCE |
 |
| GRAPH 7: PHOTOSHOP PERFORMANCE |
Ever since the dawn of desktop publishing, creatives have tended to gravitate towards the Mac. But industry-standard applications for designers and multimedia producers have long been available for Windows, too. Might creative software run better on Windows?
To find out, we started with a few tasks that will be familiar to many print professionals: we loaded a 20-page PDF into Adobe Acrobat, carried out text recognition, then exported it using Adobe’s “Mobile” optimisations. We did this using Acrobat XI Pro on OS X and Windows – this being the version currently offered as part of Adobe’s Creative Cloud service – and timed each operation.
Although the software is supposedly exactly the same on both platforms, OS X was consistently slower than Windows (see graph 6, above). The difference was a matter of seconds in this case, but applying such operations to multi-chapter publications full of hi-res photographs can take minutes or more. The cumulative time lost could cost a busy agency serious money.
Finally, we turned to arguably the most widely-used creative tool in the world – Adobe Photoshop. Again, we used the latest version – Photoshop CC – and timed a series of actions. First, we opened three 24-megapixel raw images via the Camera Raw importer. Then, we applied a 100% Smart Sharpen filter to one, with a 1-pixel radius and 10% noise reduction. We used the “Save for web” module to shrink it and export it as a medium-quality JPEG.
On the iMac, our number-crunching tasks were around 20% slower in OS X than in Windows (see graph 7, above) – an outcome fairly consistent with what we saw in Acrobat. On the MacBook Air, however, OS X managed to open the files more quickly than Windows, and raced ahead by 30 seconds when it came to Smart Sharpening. Evidently, the MacBook Air hardware allows Photoshop on OS X to take a clever shortcut that isn’t available in Windows, nor on the older iMac.
The Winner
 |
| Real-world performance varies in Adobe Creative Cloud, depending on the operating system |
Our results are a mixed bag. In some cases, such as Google Chrome testing, we’ve seen almost identical performance across Windows and OS X. Elsewhere, Acrobat XI Pro and Word 2013 for Windows made Apple’s platform appear slow. However, if you bring in a native application, the Mac looks good again.
Sadly, you can’t point to one platform and say “this one is faster”. Even though OS X and Windows run on similar hardware, performance can be different, even in areas where you might expect the platforms to be in lockstep. For the best computing experience, you can’t follow one brand. However, our results give you an idea of the strengths of some of the most widely used applications, and if you weigh this information into your buying decision, you’ll end up with the best tool for your job.
