Given that one of my main goals was to preserve the 1680-by-1050 pixel count of my 15-inch MacBook Pro, I love that I can run the 13-inch Retina MacBook Pro at that identical (though simulated) resolution. The pixels are definitely denser than on my 15-inch machine, but I've found that (even with my aging eyes) I can use it quite well: Certain websites have to be hit with Command-Plus (increase size) a few times, but I love how much more space I have when compared to using the machine in Retina mode.
When my eyes are feeling tired, I can use the in-between 1440-by-900 setting; when they're really tired, or I want to do a lot of reading or just easy Web browsing, I can set the machine back to Retina mode.
Just because it's possible, I installed Display Menu, which lets you access the full native resolution (2560 by 1600) of the display. While things were visible on the screen, unless you've got 20/10 vision, you would not want to work at this resolution for any length of time. But it was still an impressive demonstration.
One area of concern about using the "more space" modes is the warning you get when scaling resolutions in System Preferences panel, to the effect that using scaled resolution may affect performance. This only makes sense: the computer must work to create a resolution that doesn't exist natively on the display. But how much of a performance hit would I take for that?
To answer that question, and others about general performance changes over the years, I ran a number of benchmarks to test the disk, CPU, graphics, and overall performance of the new Mac as well as the two machines it'd be replacing (and, again, the iMac as a point of comparison).
I used an odd assortment of tools in my benchmarking — some of them well-known (Xbench, Cinebench), others less so (gpuTest, Unigine Valley). I don't pretend to be a professional benchmarker. I didn't test these machines in a controlled environment, and I didn't spend days at it.
Rather, I kept things simple: I created a new user on each machine (to make sure it wasn't running any of my usual add-ons), then rebooted the machine, logged into the new user account, and ran each benchmark twice, and averaged the results. I ran resolution-dependent tests (GpuTest, Unigine Valley, and portions of Xbench) at both Retina and the 1680-by-1050 resolution setting on the new Retina MacBook Pro; I also ran the GpuTest and Unigine Valley tests in an 1152-by-720 window — the largest I could fit on the 11-inch Air.
Sign up for CIO Asia eNewsletters.