And I'm saying that without taking into consideration the SSD and CPU speed the results are pretty worthless. You can't just measure performance, change multiple things, measure performance again, and then attribute the performance difference to one specific change. A meaningful comparison is hard. Why not use two machines with the same amount of RAM and compare the number of page-outs over the whole test run? Are ARM binaries on average smaller than Intel binaries? Just two things I'd look into beyond what the video does.
It is
just a video by someone. All that guy wanted to prove is, I assume, that 8GB unified memory in the M chip series pretty good at carrying heavy loads. He was actually using a M1 MBA. So, the 8GB M2 15" MBA is quite good enough for small office work, which includes some slight photo editing, web page creation and such work. The 15" M3 MBA, I suppose comes with 16GB unified memory.
That guy did a stress test with the following,
I'm not going to edit videos, open 800 photos, run Chrome web browser at all, open 16 docs each Pages and Numbers, or won't use Premiere Project, DaVinci Resolve or Photoshop, launch News, Music, or TV at the same time. Sure, I'd be opening quite a few Safari tabs, that I use an external monitor keep tract of them, sometimes few Excel sheets, some few Word docs (they can be also opened in the web in Safari), lot of work with Preview, editing images and pdfs, but won't be doing them at the same time.
At the moment, I'm using an Intel 15" MBP, (which has MS Office LTSC 2021, that will not work in a Silicon Mac) the memory pressure stays in the green all the time, more or less as below.
I'm thinking of buying a 8GB M2 15" MBA, for some unopened ones (unsold ones) are available at very good prices. My present Intel 15" MBP doesn't allow me to do that yet, as it is refusing to die. 😊
Here's the same guy
giving a torture test to his M1 MBA.