I am trying to get rid off my MacBook Pro and buy an Apple Studio computer but I haven’t decided the optimum configuration yet. I just wanted to show you guys a test I run the other day using my MacBook Pro 2018 vs my wife’s MacBook Air 2020.
MacBook Air (Late 2020)
Apple M1 3197 MHz (8 cores)
Single-core Score 2327
Multi-core Score 8145
Apple M1 (METAL) 26349
Apple M1 (OPENCL) 18891
So the first thing I noticed is that my AMD Radeon Pro 560X has a slightly lower performance than Apple M1; 26K vs 23K so I wasn’t expecting a much faster processing time of MacBook Air. CPU performance is also a slightly higher on MacBook Air but not by much.
So I processed an Olympus E-M1 Mark 2 20 MP raw image, using DeepPrime for noise reduction. Macbook Pro took 1 minute and 20 seconds and MacBook Air took only 20 seconds!!! I couldn’t believe it! It is x4 times faster.
Then I started reading about the M1 and M2 architecture and it seems that the Neural Engine is the key of this huge improvement.
So here is the question, it seems that M1 processors have 16 neural engine cores except for the ULTRA that has 32 cores. M2 seems to have 16 cores too but I don’t know if they are faster or not.
Does anybody know if M1 and M2, even having the same amount of cores, have the same neural engine performance? It seems that Memory and CPU cores is not increasing processing speed but Neural Engine does.
Yes, neural cores are the key on this kind of computing. Neural cores in M2 computers are (accordingly to Apple) 40% faster than the ones on M1 chips. Also, the Ultra version has 32 cores instead of 16, which is going to further speed up the process.
I’m sure somebody else will chime in here with some extra info and/or advise.
Consider a few aspects or real-life performance of PhotoLab:
Time it takes to render the previews, once you open a new folder in Library mode
Time it takes for the preview to completely show on switching images in Customise mode
Time it takes for the preview to react to sliders in Customise mode
Time it takes to export images
GPU power and neural engines mostly do their work while exporting…and total times for export depend on many things, e.g. the corrections you used, the number of images to be processed in parallel and how you work.
Increasing the number of images to be processed in parallel can have adverse effects, as total load shifts between processors and the bottleneck can shift to make things worse. Exporting images one by one is another thing that can spoil your happiness. Export a bunch of customised images instead…and do something else while you wait.
If ever possible, try to evaluate performance on a Mac Studio before investing, unless money is not an issue. Maybe the sweet spot is in a Mac mini with lots of memory and storage instead?
Beware of noise. All my previous iMacs were silent (except for disk chatter) and new Macs come with fans that can get noisy: https://quietmac.netlify.app
M1 2020 beats your 2018 Intel hands down, no question on that as M chip has changed computer speed radically, boot up speed and processing speed of whatever you do. so yes an M1 macbook air 8gb ram leave dust to your Intel macbook pro i7 16gb ram.
don’t get me wrong on this, 8gb ram might be minimum to run software but you want on your machine a minimum 16gb ram, crazy is you can go up to 64gb of ram on those new M2 (too much for me). why? if you going to spend that money on a computer, get something that will make it run for 8 yrs like you bought it last month! don’t think of adding RAM to your machine down the road, that’s too late and more costly. that’s both my opinion and experience. we just replaced both mine and the wife computer cause they are now 8-10yrs old i7 intel iMac 16gb ram and too old to upgrade/update software.
*** edited
don’t forget that Apple support OS for 7yrs as they introduce a new one every year! after that only Safari and *critical update are supported. so if you buy an already 2 yrs old computer… you have 5 yrs to go. is that going to be worth the money?(200-350$ cheaper), it’s up to you
You are totally right about noise, my Macbook Pro doesn’t stop making noise using the fans when I use Photolab. I am totally done with this Mac. My first computer was MacPro Early 2008 and it was far better than this laptop. The only reason why I bought a laptop was mobility and it was a terrible mistake.
I wish I could install a NPU card on a PCI Express 2.0 lane and I would continue working with my MacPro.
I’ll continue running tests but definitely these M processors are fantastic. Let’s see what I can get for a laptop from 2018.
Both Macs have 16Gb. I used to buy things to last forever but I am not totally convinced it’s the right philosophy nowadays. I would love buying an Ultra M processor with 32 neural engine cores but I rather buy a plain Max studio mac and sell it every two years. Technology changes so fast that software doesn’t run properly on old machines. Look at Photolab, it takes 1 min 20 secs!!! to process one pic on my MacBook Pro 2018!!!
I liked the idea of processing pics remotely, and that’s why I bought the laptop but the computer is super noisy, gets super hot and it’s slow. The other day when I tried the M1 Macbook Air, I was shocked! x4 times faster and computer super cool!!! If the neural engine is the most important thing, it’s going to help me out to decide what to buy next time.
