M2 MacBook Air vs 3.8GHz 8-core Intel Core i7 iMac

[Jashue]

A few days ago, I bought an M2 MacBook Air with 16GB of unified memory and a 1TB SSD. It might be the most mature computer purchase I've ever made, as I too often get bogged down in geekdom and buy machines spec'd to do things that I don't too often do. No doubt all here can relate as they look at some of the reviews for the latest MacBook Pro.

That said, I was curious to see how this new machine stacked up against my (what was then apex) 2020 i7 iMac with 32 gigs of ram.
I downloaded a 3.18GB DMG of an HD video with the intention of running the files through Handbrake on both machines under exact conditions (both after restarts and no other apps running). After hitting start on both machines, it became pretty apparent that the iMac was going to encode the file more quickly (based on anticipated ETA), but I ran with the test for a bit. After around 45 minutes, I checked in to see the progress to find that the disparity had grown to a very wide extent. The M2 laptop was as hot as a George Forman grill, and I decided to end to process to put it out of its misery.

Now— as I said before— this purchase was a responsible one. I bought it to get the kind of work done that laptops are traditionally used for, such as email, web browsing and spreadsheets etc...

BUT... the Geekbench CPU Benchmark Scores for the iMac are: single-core- 1250 and multi-core- 8161.
The Geekbench CPU Benchmark Scores for the M2 MacBook Air are single-core- 1893 and multi-core- 8739
The M2 should be a bit faster, at least before throttling. From the get-go, I could see that it wasn't even competitive.

I still love this machine. It does what I need it to do. I'm just curious.

Thoughts?

 

[1BadManVan]

Sustained heavy duty workloads was never the intention of the air so the results arent all surprising, it’s fanless after all. But for realistic workloads, it will be a fair bit quicker for the day to day work

 

[Jashue]

Sustained heavy duty workloads was never the intention of the air so the results arent all surprising, it’s fanless after all. But for realistic workloads, it will be a fair bit quicker for the day to day work

I know! But within a few minutes (before any throttling could occur) I could see that there was a huge disparity in frames per second. It doesn't really matter that much, but it was kind of surprising!

 

[Shirasaki]

I said that and I’ll say again. M1 and M2 is not all what people claim to be. They are a big jump in terms of power/efficiency from Those boring Intel processors that’s for sure, but just looking at the raw power leaves much to be desired. Compatibility with Windows (rather the lack thereof) also complicates people who needs windows regularly enough but don’t want to buy a PC.

 

[Technerd108]

The M2 MBA is a great laptop but it certainly has it's limitations. I would be curious if you had a laptop cooler or just a fan on the MBA or having it next to an open window with cool air might have made a difference?

Seems like a MacBook Pro or M2 Mini with M2 Pro might be better suited for the task?

Geekbench scores are rather subjective and test burst speed for a short time. So if you are doing tasks that are not sustained load like web browsing. Opening applications and other quick tasks it is a great measure but Geekbench doesn't show GPU performance or sustained load performance. You can try compute test on both machines to get a better idea of GPU performance which I think handbrake probably utilizes more.

Also the m series chips dip in performance when both GPU and CPU are simultaneously under heavy load and if there is no active cooling then throttling on both can occur.

The air can be heavily used but you have to do something to make up for the lack of an active cooling system. Some people have even used a frozen pad and fans with excellent results for those times when they are pushing the laptop beyond its intended use case.

Depending on the 10th gen i7 desktop processor it is not as compromised as Intel mobile is. Intel on desktop is still very competitive. Newer Intel 13th gen is faster than m series chips and coupled with a newer Nvidia GPU would be more powerful than any Mac hardware currently available if properly cooled. It can use all the watts it needs being a desktop.

Comparing a regular m2 chip that is also used in an iPad somehow doesn't seem fair against a desktop IMHO.

I am pretty sure a Mac Studio or M2 MBP would still be a lot faster than your imac.

 

[jdb8167]

I was curious to see how this new machine stacked up against my (what was then apex) 2020 i7 iMac with 32 gigs of ram.
I downloaded a 3.18GB DMG of an HD video with the intention of running the files through Handbrake on both machines under exact conditions (both after restarts and no other apps running).

It’s software so there is going to be different levels of optimization between different architectures. Handbrake has a large number of options and you didn’t give any details of how you did the compression.

The x86-64 CPU could have been using AVX while the M2 was not using the similar NEON SIMD instructions. More details would help figure out why the M2 was slower than expected. I would expect newer codecs to be better on the M2 than older. So try with HEVC instead of h.264 for example.

Apple has their VideoToolbox for hardware assisted video compression. That should work for both the i7 QuickSync and Apple’s own Media Engine. The downside of using hardware compression is that to get similar quality as from the software compression the file sizes are generally larger.

 

[ElectricPotato]

That is a worthwhile data point and a good reality check. Too many people buy into the benchmark hype. A MB Air is not a supercomputer or even a desktop, regardless of processor. The laws of physics apply to all chip makers.