Both Microsoft and Apple claim their new operating systems are sleeker and faster than ever before. But OS upgrades are often fraught with trouble. We decided to pit these two champion OSs against each other to see which would perform an in-place upgrade faster.
The big challenge with this sort of test is making sure the testing platforms are close enough. There's always a bit of a spec war when pitting a Mac against a PC. This is because while both use common components, there's inevitably something different.
Not this time. This time we took a 2012 Mac mini with two SSD drives, and we upgraded an in-place version of OS X on one drive, and an in-place Boot Camp version of Windows on the other drive. This means that our network configurations, data channels, and disk performance were absolutely identical. This machine includes native SSDs purchased with the machine from Apple, as compared to my many aftermarket Mac mini SSD installs.
The starting point from an OS perspective was not identical, however. We started on the Windows side with Windows 8.1, which had received all available Windows updates just a few days earlier. It was about as current for 8.1 as it gets.
On the OS X side, we started with OS X 10.8.4, better known as Mountain Lion. As I discussed in If it ain't broke, break it: Why putting off software upgrades can make things worse, Google services no longer support Mountain Lion. That puts it three releases behind the current El Capitan release of OS X.
Having followed Ed Bott's guidance on preventing automatic Windows 10 downloads and installs (see How to block Windows 10 upgrades on your business network and at home, too), I didn't have a pre-existing Windows 10 installer on my machine.
Before I started the clock, I pointed my browser to Microsoft's Get Windows 10 page, and downloaded the starter installer. That only took a minute or so, and put both upgrades on equal footing.
There are two key phases for these upgrades: the download and the install. In terms of download, it should be noted that I have a relatively fast Internet connection. According to Netflix' new Internet speed test service, I get between 140 and 150 Mbps download speed. Obviously, if you have faster or slower bandwidth, your OS download performance will differ.
The second phase is the install. In my case, I did two in-place upgrades, which I don't necessarily recommend for the best possible results. I tend to try in-place upgrades first, especially since I have very fully-configured machines. But it's long been IT doctrine that fresh installs are the best approach. I've generally been pretty lucky with in-place upgrades (as opposed to completely fresh installs), but you do take your chances.
[We interrupt this article with an important best practice message]
Before I go on to discuss benchmark results, it's important to remind you to backup your systems before you do an upgrade like this. OS upgrades often fail. No matter which vendor fails, it's a sad day when your upgrade takes your machine and your data down with it. Always do a thorough backup before upgrading.
[And now, back to benchmark results]
I started the OS X El Capitan install at 3:15pm, and successfully booted into an El Capitan desktop at 3:48pm. The total process of both install and download took 33 minutes. Download was 14 minutes and the upgrade took 19 minutes to complete.
On the Windows side, I started the Windows 10 install at 11:25am. The download completed at 12:01pm, having taken a total of 36 minutes to complete. The upgrade itself (which crashed twice), took an 102 minutes to complete (including the power cycling for the crashes). Overall, the Windows 10 upgrade took 138 minutes - 105 minutes more than OS X.
There are a few things to consider. First, although I have had great success wth running Windows natively on Mac minis, this is a Mac. There could be some issues associated with how long it takes Windows to be comfortable with hipster hardware.
Second, the crashes were unfortunate. One simply blanked my screen and stayed that way. While this could be attributed to the Mac environment as well, I've definitely had crashes as part of my other Windows 10 upgrades. In fact, the only machine on which I haven't had a crash as part of the Windows 10 upgrade was another Mac mini.
While the OS X upgrade was both painless and fast, that's not always the case with OS X upgrades. I went through hell upgrading to Mavericks back in the day. I even wrote an open letter to Tim Cook complaining about it. El Capitan also gave me fits until it got a point upgrade from Apple.
On the Windows side, a little over two hours for a Windows upgrade is pretty good. I've been quite satisfied with Windows 10. While, like my ZDNet buddy Adrian Kingsley-Hughes, I've been reducing my Windows headcount, I still have production work-related needs for Windows machines in my shop. By July, most will be running Windows 10.
Adrian also tells me that after you do the upgrade, if you had any open programs, OS X will reopen them. Personally, I find that unnatural and disturbing, like when Alexa chimes in when my wife and I are having a heated discussion. It just seems wrong. I don't ever let the Mac reopen my applications after reboot. Reopening them after an upgrade is disturbing at some kind of primal level.
Windows doesn't try to perform any unnatural-seeming parlor tricks (unless you consider Cortana, but I still primarily think of her as the slightly twisted friend who'll keep you alive when battling the Covenant). That said, ZDNet's Jack Schofield pointed out that if you do have applications that were in your Startup folder, they will start back up when your upgrade is finished. That doesn't seem nearly as unnatural as it does in OS X because, well, you put them there in the first place.
Of the two, OS X and Windows 10, which upgrade went faster? Clearly, by a 4.1-to-1 margin, the OS X upgrade was faster. It was also smoother, requiring no power cycles or forced reboots, and it had no crashes. But both are fine operating systems. Once you get them upgraded to their latest versions, regardless of which you use most, you should do just fine.