There are a few ways one might measure the performance of a laptop. You can run benchmarks. Or you can just use the thing.
I’ve decided to try both approaches with the new MacBook Air. After all, performance is one of the major concerns I’ve had with this purchase. This needs to be explored, and it needs to be explained.
On that note, I’ve followed the comments to my Apple MacBook Air (2018) First Impressions article with great interest. This isn’t unique to Thurrott.com, of course, but comments on the web often devolve into rank, partisan arguments, largely from people whose minds are already made up. And, for sure, there’s been some of that. As expected: Apple sits at a nice nexus for the faux outrage crowd, thanks to its high prices, curious component choices, and corporate ego. The company just makes it too easy sometimes.
But I’d like to give the readers of this site some credit, too. This could change as the digital cockroaches wake up this morning and realize that the Windows Boy ™ had the temerity to write about an Apple product in a reasonable way. But there were more thoughtful comments to that article than bullying or pointless anti-Apple comments. That’s good news, and appreciated.
The legit comments seem to boil down largely to a few main topics: Price, performance, and the durability and reliability of the MacBook Air’s controversial third-generation “butterfly” keyboard.
I’ll keep a nervous eye on that last one, of course. But this can only go one of two ways in the short-term, during the two or three weeks in which I’ll evaluate this machine and then write a review. The keyboard will fail, forcing me to give up on it and return it to Apple, as I had to do recently with the terribly-made Pixel 3 and Pixel 3 XL (with Google). Or the keyboard will simply work fine during that time frame. Which will prove nothing.
Put simply, unless I do experience a major fail here, there’s no way I can opine on keyboard reliability. And so all I can do is wait. And hope. And wonder.
But I can measure performance. And this one I really do want to figure out. It’s tied to what I feel is the most curious component choice that Apple made with this machine, its dual-core Y-series Intel processor.
Everyone—from seasoned industry analysts down to the comment lurker trolls that dog all websites these days—has an opinion about this choice. On the outside—e.g. everyone who has never actually used a new MacBook Air—those opinions are generally very negative. After all, Apple is foisting a less-powerful CPU on its users in an age of quad-core laptops. And it’s then charging a heady premium for the privilege.
That’s a fair argument in theory, and it’s certainly one I’ve made myself. But I suspect you’d find a contrary view if you were to tally the performance ratings in the reviews that have been published so far. In other words, I believe that no one who has actually used a new MacBook Air has complained about the performance.
This can be rationalized in a few different ways, of course. I’ve written and spoken about the weird effect where one’s expensive technology purchases reflect their decision-making capabilities and thus become personal statements of worth. And that these people will defend the products they buy, and the decisions they make, in order to save their egos.
There’s also a perceived Apple bias, both in the mainstream press and in the blogosphere. And it’s not hard for this audience to rationalize that Apple “needed” to deliver a product here that fit neatly into its modern Mac lineup. That a dual-core part was somehow the only choice that made sense.
Whatever. This is where I’m at.
In theory, Apple’s decision to use a dual-core Y-series Intel CPU is, of course, outrageous. The industry has shifted to quad-core U-series Intel processors, did, in fact, make that transition a year ago. And the Windows PCs that the MacBook Air compete with are all using these CPUs. And are benefiting from the additional performance they provide, and are doing so without any hit to battery life.
But Apple did choose a Y-series part. Whether it did so because of the thermal requirements of the new MacBook Air chassis, or because it had really been waiting for some 10 nm part that is now delayed indefinitely is sort of beside the point. This is the reality: The MacBook Air stands alone with this curious 7-watt Y-series CPU.
And … in my admittedly limited experience so far, it’s worked fine. I’ve noticed no performance issues at all, across applications like Google Chrome, with multiple tabs, Adobe Photoshop Elements 2018, Skype, MacDown (the Markdown Editor I used to write that first article), or anything else I threw at it.
Apple used the term “everyday tasks” to describe the sorts of work that a typical MacBook Air user might undertake with this machine. But that may be underselling it. I think the new MacBook Air is a capable performer in what I’d call “traditional productivity tasks,” which includes email and web browsing, of course, but also Microsoft Office and, yes, even some Photoshop.
I’ll also see how some more advanced applications, like Apple’s Xcode development environment, perform. But Xcode worked/works just fine on my 2014-era MacBook Air, and I have little doubt it will work similarly on this one. Maybe macOS is just more efficient than Windows, folks.
Beyond this anecdotal usage data, which will turn into “actual, real-world experience” over time, it’s also helpful to look at benchmarks. Not in isolation. But in context.
Geekbench reports that the new MacBook Air and its Intel Core i5-8210Y processor score 4248/7828 (single-core/multi-core). That’s for a MacBook Air with 16 GB of RAM, just like the one I bought.
Those scores are higher than the 3335/6119 posted by the previous-generation MacBook Air. And they’re lower than the 4314/9071 posted by the non-Touch Bar-based MacBook Pro, which is the next Mac up the org chart. So it looks like some of that rationalization makes sense: The new MacBook Air landed right where it needed to in the lineup. That the old MacBook Air is $200 less expensive than the new model, and that the MacBook Pro is $200 more expensive, is particularly interesting.
I also ran my standard performance test, in which I encode the 4K video Tears of Steel to a high-quality 1080p version using Handbrake. Here, the MacBook Air didn’t do so well, posting a time of 1:52. And there was loud and constant fan noise during the entire procedure.
Let’s look at this one in context, too.
By comparison, my 2014 MacBook Air—with its 1.4 GHz Intel Core i5-4260U processor and 8 GB of RAM—needed 2:40 to complete this encoding test. So one might argue that the new machine is indeed a worthwhile upgrade.
But Surface Laptop 2—with a much newer quad-core 1.7 GHz Intel Core i5-8250U processor and 8 GB of RAM—needed only 51 minutes to complete the encoding, an excellent score. Here, the new MacBook Air does not compare favorably with a modern Windows laptop, and arguably the PC with which it most closely competes.
On a side note, both the 2014 MacBook Air and Surface Laptop 2 issued a steady fan noise during their respective encoding tests. But the new MacBook Air’s fan was far louder than either. I felt like I was abusing it somewhat just running this test. That said, the new MacBook Air remained relatively cool to the touch, and on par with the Surface Laptop 2. (The older Air heated up more.)
Looking over the results of the many PCs I’ve reviewed, I can see that the new MacBook Air falls right at the back of the pack when it comes to this encoding test. Modern, quad-core laptops typically need a bit over an hour to make this conversion. So the MacBook Air needs about twice as much time as the average these days. But the lowly HP Envy x2 (Intel) was even worse. It also utilizes a Y-series processor, but it was held back by its paltry 4 GB of RAM and was even slower: This terrible PC needed almost three hours to finish this test. That’s even worse than a 2014 MacBook Air.
Look, I’ll keep using the new MacBook Air. But my gut feeling is that it does indeed deliver on the performance one would expect of this laptop when used at the “everyday tasks” for which it was designed. That’s a nice surprise, and a useful reminder that going into something like this with preconceived notions isn’t perhaps the best approach.