Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Pugly

macrumors 6502
Original poster
Jun 7, 2016
411
403
I just bought a new MBA (i5/16GB/256). I ran some tests in Logic and Cinebench to compare it with my old Air, and thought some here might be interested in what I found.

With the talk of throttling, and cooling limitations I wanted to see how it compares to the old Airs. If this new Air isn't fast enough I would return it. I have the most powerful 2015 MBA, an i7 with 8GB. Going off Geekbench scores, it seems like this will perform about as good as the i3 2020. I was worried that I would get a computer that's only the same speed under sustained loads.

Anyways, from the test the 2020 Air is about 50% faster in sustained heavy workloads. It seems like the Air doesn't like to run the fan unless it absolutely has to, so you can boost things slightly by cranking the fans to 8000rpms at the start. It rarely went up to the full 8000rpm though, and usually stuck around 6500. Curiously this was the top speed of the 2015 fans too. The fan is less annoying sounding than the old Air too, though at 8000rpm it can get louder too.

In Geekbench 4 scores the 2020 got a score of ~3000 compared to the old one's ~1500.

My main use for this Mac is Logic, and I'm pretty familiar with the performance I got out of the 2015. I was hoping for double the performance, but alas it seems it's only about 50% better. I'm thinking more cores will be better overall though, because audio tasks are more multithreaded than single threaded.

For my tests, I used the same plugin preset and settings for each with the same midi region. The audio buffer was kept at the default 256 samples. I kept adding tracks and played until I ran into a CPU warning. I then took away tracks and played for a couple of minutes until I would get no warnings. I used the most demanding plugins I could. I closed all the other apps.

Plugin2015 i72020 i5
Bazille with a polyphonic part14
MODO DRUM46
chipsynth PortaFM1929
Satin (tape effect)2743

I also wanted to see what I would get in Cinebench r20. For added comparison I also ran the tests with the fans going full blast from the start.

2015 i72020 i5
Auto fans684979
Full blast fans6981038

So about 45% better. The boost from using the fans is pretty significant I think. It shows the fans are at least doing something. I think this machine is a hybrid of 12" MacBook and the old Airs. It's designed to stay quiet as much as possible and run up the cpu temps, but it has a fan to cool itself if it absolutely needs to.

Overall I'm happy with the performance. It's not the double I was hoping for, but still noticeably better. Anyone looking to do some moderate tasks shouldn't worry about this computer.
 
Thank you for the review.
Do you use an external or internal audio card?

These tests are all with internal audio.

I don't notice any difference in performance using an external audio card. Though I'm not using any fancy multi input/output setups.

A couple other observations, I started the tests using an external 1080P screen on the new Air, that did make the top performance worse(or I was getting about the same performance which was alarming). The benefits showed once I ran off the internal screen. One good thing about the crappy screen of the old Air is the graphics didn't have to work as hard just to render the basic desktop, which gave it a slight benefit sometimes over a comparable Pro.

Also the quad core 2020 Air only used 7 threads instead of the full 8 in Logic, as reported by Logic's CPU meter. That might be by design though, to give the OS 1 thread to do basic stuff in the background. The old Air used all 4 threads pushed to the max. So maybe there is still a little more headroom than what I benchmarked on the 2020 Air.

And I think the old Air is slightly faster per core when using all the cores. The amazing single core performance of the new Airs goes way down when you need all of them cranking. It's still enough to give it an edge because the 2020 has more, but just something interesting to see.

I think in more normal situations, when it's not being cranked to full blast using these demanding plugins, it'll be better though. Since it'll run cooler spreading the cpu to more cores at a slower speed, than pushing a few cores to a higher speed. Running a bunch of EQs and Compressors is no where near as hard as playing a polyphonic part through a demanding instrument.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.