M2 Pro and M2 Max

1674254021698.png

Huh...that's modern tech journalism folks!
 
Has anybody checked on trade in allowances from Apple.....

Parking aside the sanity of even looking to trade in a 16" M1 Max with 64GB ram and 4TB storage......

This was a laptop that cost me over CAD $6300 in 2021.

Fast forward to today.... apple will offer me $630 for the same laptop.

I'm pretty sure when I traded my 16" i9 intel MacBook Pro for the M1 Max, they offered me a significantly higher trade in discount relative to the original purchase price.

I'm shocked - literally shocked at just how crappy the trade in allowance is from Apple.
What can they do with your used laptop if it's anything other than pristine? They are paying you a fee on top of paying to dispose of your computer in an environmentally friendly way. The only machines that get sold on Certified Refurb are ones that get returned within the 14 day period. There's no profit motive for them to build out a massive logistics chain to refurb old models that will just undercut their new product lines.
 
Courtesy of Apple Insider, they supplied a handy chart showing relative Geekbench performance among M1 and M2 series Metal performance.

Screenshot 2023-01-20 at 6.25.31 PM.jpg


My takeaway is that a theoretical M2 Ultra is going to be close to a W6800X in performance. If that is where the Mac Pro lands, then that won't please the blistering performance crowd, but may be satisfactory enough for Apple to declare that a win. I wish I had the exact quote, but I believe Apple has said that the most popular GPU shipped inside the 2019 Mac Pro is the W5700X. An M2 Ultra will have no issue outclassing that. Also, assuming that the Apple Silicon Mac Pro ships in March, that's well before the launch of AMD's RDNA3 7000-series professional cards. It's not what would have been possible with an "Extreme", but perhaps close enough that Apple feels it will be fine with an M2 Ultra, and won't need third-party GPUs.

If this comes to pass, there's going to be an intense uproar among enthusiasts who own Mac Pros to tinker with, but they are likely noisier than their purchasing power. It's like the PC hotrod gamer crowd, who keep declaring Apple dead and that Tim Cook should be fired, even though the Mac is the only traditional computer platform that isn't down double digits in shipments, and is increasing in market share. I suppose we shall see, sooner or later.
 
Courtesy of Apple Insider, they supplied a handy chart showing relative Geekbench performance among M1 and M2 series Metal performance.

View attachment 21401

My takeaway is that a theoretical M2 Ultra is going to be close to a W6800X in performance. If that is where the Mac Pro lands, then that won't please the blistering performance crowd, but may be satisfactory enough for Apple to declare that a win. I wish I had the exact quote, but I believe Apple has said that the most popular GPU shipped inside the 2019 Mac Pro is the W5700X. An M2 Ultra will have no issue outclassing that. Also, assuming that the Apple Silicon Mac Pro ships in March, that's well before the launch of AMD's RDNA3 7000-series professional cards. It's not what would have been possible with an "Extreme", but perhaps close enough that Apple feels it will be fine with an M2 Ultra, and won't need third-party GPUs.

If this comes to pass, there's going to be an intense uproar among enthusiasts who own Mac Pros to tinker with, but they are likely noisier than their purchasing power. It's like the PC hotrod gamer crowd, who keep declaring Apple dead and that Tim Cook should be fired, even though the Mac is the only traditional computer platform that isn't down double digits in shipments, and is increasing in market share. I suppose we shall see, sooner or later.

i wish they graphed M1 and M2 as separate lines to better show the scaling of each. Someone get on that.
 
Courtesy of Apple Insider, they supplied a handy chart showing relative Geekbench performance among M1 and M2 series Metal performance.

View attachment 21401

My takeaway is that a theoretical M2 Ultra is going to be close to a W6800X in performance. If that is where the Mac Pro lands, then that won't please the blistering performance crowd, but may be satisfactory enough for Apple to declare that a win. I wish I had the exact quote, but I believe Apple has said that the most popular GPU shipped inside the 2019 Mac Pro is the W5700X. An M2 Ultra will have no issue outclassing that. Also, assuming that the Apple Silicon Mac Pro ships in March, that's well before the launch of AMD's RDNA3 7000-series professional cards. It's not what would have been possible with an "Extreme", but perhaps close enough that Apple feels it will be fine with an M2 Ultra, and won't need third-party GPUs.

If this comes to pass, there's going to be an intense uproar among enthusiasts who own Mac Pros to tinker with, but they are likely noisier than their purchasing power. It's like the PC hotrod gamer crowd, who keep declaring Apple dead and that Tim Cook should be fired, even though the Mac is the only traditional computer platform that isn't down double digits in shipments, and is increasing in market share. I suppose we shall see, sooner or later.
It looks like the M2 Max outclasses the W5700X no?

Also, it could be just me, but that chart makes the gap to the M2 Max from the Pro look pretty good.
 
It looks like the M2 Max outclasses the W5700X no?

Also, it could be just me, but that chart makes the gap to the M2 Max from the Pro look pretty good.

Looks like the M2 Max is around 65% than the M2 Pro. The M1 Max is around 54% better than the M1 Pro?
 
It looks like the M2 Max outclasses the W5700X no?

Also, it could be just me, but that chart makes the gap to the M2 Max from the Pro look pretty good.

I'd also say it's worth bearing in mind that, at least for the M1 generation, Geekbench Metal scores were relatively low compared to how the GPU performed in most tasks when comparing against AMD Macs.
 
Here is my highly technical take: my next Mac is going to make my Intel Mac mini look really stoopid.

Just for laughs, let us compare the Geekbench results between the latest, to the not so greatest. First, here are Geekbench scores for the M2 Pro Mac mini:

m2-pro-geekbench.jpeg


Now, I just ran the same test on my 4-core, non-HT, 3.6Ghz, Core i3 2018 Mac mini, and got this:

i3Macmini.jpg


As I mentioned earlier, the RX 580 eGPU I have attached gets the same Metal score as the M2 Pro. Of course, that AMD GPU is kept inside of a graphite-cooled Soviet nuclear reactor.

I don't know whether I'll be upgrading my Mac mini to an M3 or M4 generation, I'll need to see the A17, M3, and know when Apple finally drops macOS support for Intel. However, when I do upgrade to a new machine, it's going to blow this old dog out of the water, it won't even be close.
 
Here is my highly technical take: my next Mac is going to make my Intel Mac mini look really stoopid.

Just for laughs, let us compare the Geekbench results between the latest, to the not so greatest. First, here are Geekbench scores for the M2 Pro Mac mini:

View attachment 21402

Now, I just ran the same test on my 4-core, non-HT, 3.6Ghz, Core i3 2018 Mac mini, and got this:

View attachment 21403

As I mentioned earlier, the RX 580 eGPU I have attached gets the same Metal score as the M2 Pro. Of course, that AMD GPU is kept inside of a graphite-cooled Soviet nuclear reactor.

I don't know whether I'll be upgrading my Mac mini to an M3 or M4 generation, I'll need to see the A17, M3, and know when Apple finally drops macOS support for Intel. However, when I do upgrade to a new machine, it's going to blow this old dog out of the water, it won't even be close.

My random assumption is that Apple will a settle for around 20% improvement (M3* vs M2*), since that seems to be, more-or-less, what they do every year since the A5. That means the M3 would be clocked lower than it could, in exchange for a large (10-15%) power reduction. But, for desktops, and for laptops in high performance mode, they may let the clock loose, and you’ll see a bigger year-over-year improvement than is typical (30%).

On the GPU side, I expect ray tracing, and I don’t know if that effort means that you don’t see a bigger than normal GPU performance improvement (in other words, they may be so focussed on getting that working that they aren’t worried about performance as much as usual).

I’m quite sure I will be getting an M3 MBP, but mostly for improvements like OLED screen, wifi 7, TB 5, or whatever. I can’t come close to the performance boundaries of my M1 MBP now that I am not running massive design automation jobs anymore.
 
🤦‍♂️

Just out of morbid curiosity … who?

I’m hoping that it’s more a typo than anything else? Although that would still be a comment on why copy editors are important even though it’s virtually impossible to have one anymore.
I should have definitely included a link. It's from Macworld...lol. I agree it's probably a typo, and I realise it's tough gig. It just made me laugh coming from Macworld.
 
Yeah, I get that.

I guess when I see something like a Mac Pro what races through my mind is what could someone do with that, that's off the beaten path and can take advantage of the supporting infrastructure (slots/bus, beefy power supply, high speed ethernet communications to whatever, support for high-res displays, high performance CPU, etc) all in a nice and robust chassis that I could never build on my own.

It's just begging to be a product used in some other interesting context. :)

Livestreaming synthesized arts and entertainment... high density orchestration. Bruckner... :ROFLMAO:
 
I'm kind of fascinated with the idea that the M2 was a bad update but the M2 Pro is good. I see a 48% improvement from the M1 to the M2 on Metal compute. My M1 8-core GPU MacBook Air gets about 20700 and the M2 10-core GPU MacBook Air gets about 30600. Single core CPU is up about 12% and multi core is up about 19%. These numbers compare pretty well with the M2 Pro upgrade but were considered underwhelming.
Hm, you're right. Probably a combination of unrealistic expectations. The M2 took a looong time to ship since the release of the A15, there were even rumors of it being A16 based...

i wish they graphed M1 and M2 as separate lines to better show the scaling of each. Someone get on that.

You mean like this?
Screenshot 2023-01-21 at 03.14.08.png
 
the RX 580 eGPU I have attached gets the same Metal score as the M2 Pro
Keep in mind the TB connection limits the full performance of RX 580 and the fact its made on an Global foundry node 14nm. It was power hungry/not as efficient when compared to Nvidias GTX 1060/70/80 series cards.

I would compare it the latest Nvidia GPUs and see how far behind Apple is in Hardware based RT and compute.
 
My guess is that the Mac Studio won't be updated until after the Mac Pro ships, since Apple would want it to be the king of kings among all Macs. The Mac Studio may not get updated with a the M2 generation, or when it does receive an update, the M(x) Ultra will be missing, saved exclusively for the Mac Pro.

Why limit the Mn Ultra to the Mac Pro...? I am sure there are many who need the horsepower of the Mn Ultra SoC, but do not need PCIe slots, but you want to force said slots and the extra cost involved on the end user for imaginary product segmentation...?

And if Apple were going to reserve the Mn Ultra & Extreme SoCs to the Mac Pro, then the Mac mini would have stuck with just the Mn base SoC, keeping the Mn Pro & Max SoCs to the Mac Studio...?
 
This shows how the Geekbench compute scaling compares for the NVIDIA RTX and Apple M1 series. NVIDIA shows about 80% scaling across the entire range—i.e., if the TFLOPs is Y times as large, GPU compute will be ~0.8 Y times as large. For the M1, it's 97% between the M1 and M1 Pro, dropping to 73% between the M1 Max and M1 Ultra.

[Edit: In case there's some confusion about the purpose of this, let me clarify: I was curious whether the drop-off in scaling that occurs with the higher-end AS GPUs also occurs with NVIDIA. From the graph, it doesn't seem to. Also, the reason I'm using last-gen rather than current-gen GPU's is because these series are more complete, and thus have a wider range of GPU's (good if you're looking at scaling behavior), and because these are the ones for which established scores are conveniently available.]


1674291177899.png

Formula to compute TFLOPS:

ALUs x (1 scalar FP32 instruction)/(ALU x cycle) x frequency in cycles/second x 2 FP32 FMA operations/(scalar FP32 instruction) = FP32 FMA TOPS

Source of ALUs and frequencies for NVIDIA:


1674291424633.png
 
Last edited:
Back
Top