M2 Ultra Chip Benchmark Results Reveal Impressive Performance Gains

[Alan Wynn]

And yes, Professionals 100 percent care about repairability.

Nope. Having been responsible for technology at two Film Studios and quite a few visual effects and post houses, no one cared about that at all. Machines were either leased or purchased with service contracts, and we would never repair them ourselves. All the Macs had AppleCare/AppleCare+ and it was Apple’s problem to repair them.

A component goes down, it's a matter of less money and time to replace the part and you are up and running again, instead of tossing your system and getting another.

That just is not how real professionals work. They purchase or lease machines with support. No one who is getting paid to use a computer can afford to sit around while they diagnose a problem with a machine. That is someone else’s problem.

 

[chucker23n1]

Nope. Having been responsible for technology at two Film Studios and quite a few visual effects and post houses, no one cared about that at all. Machines were either leased or purchased with service contracts, and we would never repair them ourselves. All the Macs had AppleCare/AppleCare+ and it was Apple’s problem to repair them.


That just is not how real professionals work. They purchase or lease machines with support. No one who is getting paid to use a computer can afford to sit around while they diagnose a problem with a machine. That is someone else’s problem.

This. Companies don't have the time and expertise to repair stuff. Nor do they want the liability.

 

[The point and shoot pro]

Not what I am talking about. but ok!

 

[lhammer610]

Keep dreaming . https://www.cpubenchmark.net/singleThread.html

In terms of power consumption yes, but it is just thanks to TSMC more than apple.

While TSMC is a big part in the reduction in power consumption, the ARM architecture is also crucial.

I keep wondering when Intel is going to start licensing ARM or develop their own variation.

Thanks for the link to the benchmarks. I don't see a link to multithread performance.

 

[NT1440]

Export after completed editing.

And yes, Professionals 100 percent care about repairability. A component goes down, it's a matter of less money and time to replace the part and you are up and running again, instead of tossing your system and getting another. A drive craps the bed in my systems, I get another drive and in a couple of hours my system is right back where I was. If it happens with a mac, it's oh to the scrap heap with you and buy another. Champions for the environment? Not even close, but thats a different topic.

If your self employed that’s how it works.

Most people aren’t self employed and have IT teams.

I worked at ESPN, when a machine went down we went to the build lab, got the replacement, and deployed it.

Same at every company and university I’ve worked for in nearly 15 years of IT work.

The idea that the *user* of the computer is the same person repairing it is…not typical in the business world.

 

[chucker23n1]

While TSMC is a big part in the reduction in power consumption, the ARM architecture is also crucial.

I keep wondering when Intel is going to start licensing ARM or develop their own variation.

They used to.

StrongARM - Wikipedia

en.wikipedia.org

XScale - Wikipedia

en.wikipedia.org

Intel has had all kinds of architectures, including the i960 and later Itanium, but ultimately kept going back to x86.

Keep dreaming . https://www.cpubenchmark.net/singleThread.html

In terms of power consumption yes, but it is just thanks to TSMC more than apple.

Nah, Apple's design plays a huge role in efficiency.

 

[Morgonaut]

I really think the Mac community should stop trying to define and use the word "professional"

Completely agree. Most people don't realize professional means doing something for living, nothing less nothing more. You can have awesome experts in some area which are doing it just for fun which would exceed the box called professional many many times or on the other side professionals which are doing it for living but lack even elementary knowledge about their topic.

 

[Morgonaut]

I think it is safe to say that the Mac Pro will never see higher clock speeds than the Mac Studio, except for some exceptional corner cases where the Mac Studio might thermally limit a bit sooner.

I did in last 2 weeks so many real life tests and despite I was able to push the Mac Studio M2 Ultra maxed out to 105C temp it still did not throttle at all

 

[maccan]

I did in last 2 weeks so many real life tests and despite I was able to push the Mac Studio M2 Ultra maxed out to 105C temp it still did not throttle at all

Is this not beyond max temp specs?
Did you perform a very long task using all cores and quite high ambient temp?

 

[ZombiePhysicist]

 

[DavidSchaub]

Re: LTT's video

It is such a good example of LTT videos covering Apple stuff.

Are Apple's bar graphs almost entirely useless, yes.

Does Apple compare new products to ancient Apple products, not their competitors, yes.

Does the LTT video say anything incorrect, I don't think so. All seemed pretty accurate and clear.

That being said, Apple's web page is not trying to sell these things to PC users, they are marketing to Intel Mac users, so comparisons to ancient slow Intel Macs make sense.

Really, all Linus is saying is: Apple please stop cherry picking performance benchmarks and labelling graphs so badly. Which is reasonable.

The real question is: How many people are so platform agnostic that they need a comparison between Apple's current stuff VS x86 stuff?

I kinda love the videos, so I hope LTT keeps making them

 

[Morgonaut]

Yay, its faster than a MacOS system based on a high-end 2019 intel-CPU. But time has passed ...

surprisingly after I spent several weeks comparing all new M2 AS Macs in real life tests I will probably go back to my 2019 Mac Pro because it's still freaking good compared even to M2 Ultra

 

[Morgonaut]

Has anyone found benchmarks with the M2 Max Studio 38 core gpu vs. M2 Ultra 60 core

Or the M2 Ultra 60 core gpu vs the 76 core?

Seems like all the M2 Ultra test units are all the 76 core gpu.

My napkin math is guessing the 60 core gpu is about 165K on Geekbench 6 metal, vs.. 210k for the 76 core. ~ 27% which is exactly the core count ratio.

I tested deeply Mac Studio M2 Mac base model versus Mac Studio M2 Ultra, surprisingly it is scaling well unlike the M1 Ultra, but it depends on applications, well written ones will benefit

 

[Morgonaut]

I’ve also looking for the 60-core GPU vs 76-core GPU comparison, only seeing the 76-core benchmarks. Would like to know if it’s worth $1000 more!

it only depends on you, from my experience Apple Silicon is still underpowered on GPU so each GPU core more counts. In my workflow the GPU is limiting factor and I see all the time 100% utilization of GPU. I would need double/triple that amount of GPU cores in my system = 2x76 or more would be great

 

[Morgonaut]

I kinda love the videos, so I hope LTT keeps making them

I hope LTT will not keep making them anymore, because the value is zero - they focus on stupid things like gaming frame rate etc. which Mac user would not care. LTT have their Mac dedicated channel where the information value is a bit better, but still, they are PC guys so videos about Mac from PC perspective is shifted...
Just because some application is running native as Apple silicon one that still does not mean its optimized for that platform. So comparing for example Blender on AS and on PC with CUDA is insane when it had tons of years of development against almost no development for AS.
People should compare it for example on video production where DaVinci Resolve is truly optimized for PC as well as AS. And that's where AS shines, there is not a single PC even with 13900k and 4090 which would win this round...

Last but not least people should use the right tool for the job done, so there are areas where PC would be better idea to use, on the other hand many areas where people would benefit from having a Mac.

 

[Morgonaut]

Is this not beyond max temp specs?
Did you perform a very long task using all cores and quite high ambient temp?

Nobody knows what Apple consider a safe temp or not. Obviously I don't feel comfortable to run it so hot, we all know dust and heat are computer's biggest enemies (when I don't count the user) 😂

 

[maccan]

Nobody knows what Apple consider a safe temp or not. Obviously I don't feel comfortable to run it so hot, we all know dust and heat are computer's biggest enemies (when I don't count the user) 😂

According to https://www.cpu-monkey.com/en/cpu-apple_m2_ultra_60_gpu, Tjunction max = 100°C.
The machine should be able to run the highest load as long as it is necessary without overheating. Is this maybe where the difference to the Mac Pro 2023 comes in?
On the 2019 Mac Pro (12 Core) I can run 24 Threads at 100% without any time limit (some scientific MC calculations), sometimes running 3-4 days nonstop. Never observed temperatures higher than 90°C even during hot ambient temps in summer. Fans are ramping up somewhat to keep it at this temp. Wonder why in your case the fans do no kick in, even at > 100°C. Thought the new M2 chips are much more power efficient and cooling would never be a problem. I think stressing the chip for a long period of time with > 100°C will shorten its lifetime dramatically.

 

[Morgonaut]

According to https://www.cpu-monkey.com/en/cpu-apple_m2_ultra_60_gpu, Tjunction max = 100°C.
The machine should be able to run the highest load as long as it is necessary without overheating. Is this maybe where the difference to the Mac Pro 2023 comes in?
On the 2019 Mac Pro (12 Core) I can run 24 Threads at 100% without any time limit (some scientific MC calculations), sometimes running 3-4 days nonstop. Never observed temperatures higher than 90°C even during hot ambient temps in summer. Fans are ramping up somewhat to keep it at this temp. Wonder why in your case the fans do no kick in, even at > 100°C. Thought the new M2 chips are much more power efficient and cooling would never be a problem. I think stressing the chip for a long period of time with > 100°C will shorten its lifetime dramatically.

Darling I would not trust any "cpu monkey" And even if I would my temps are still always higher than the Tjunction, same with M2 Air, M1 Max MBP etc. And BTW your W3235 Tcase temp is 72C so if you run it constantly at 90 then good luck. At least you can replace it in case it will fail unlike on AS

 

[maccan]

Darling I would not trust any "cpu monkey" And even if I would my temps are still always higher than the Tjunction, same with M2 Air, M1 Max MBP etc. And BTW your W3235 Tcase temp is 72C so if you run it constantly at 90 then good luck. At least you can replace it in case it will fail unlike on AS

Tcase is not the same as Tjunction!

 

[Morgonaut]

Tcase is not the same as Tjunction!

That's correct, I never said it is the same!

 

[maccan]

That's correct, I never said it is the same!

Had the impression you messed up these terms since you were stating that Tjunction 90°C would be problematic. I was referring to Tjunction not Tcase!

 

[Morgonaut]

Had the impression you messed up these terms since you were stating that Tjunction 90°C would be problematic. I was referring to Tjunction not Tcase!

Uff, I'm not an idiot, I know very well the difference. I'm so sorry I tried to help! I will not make the same mistake anymore.

 

[maccan]

Uff, I'm not an idiot, I know very well the difference. I'm so sorry I tried to help! I will not make the same mistake anymore.

Thank you!

 

[Morgonaut]

Thank you!

Are you kidding me? So now you really pissed me off!

 

[Alan Wynn]

it only depends on you, from my experience Apple Silicon is still underpowered on GPU so each GPU core more counts. In my workflow the GPU is limiting factor and I see all the time 100% utilization of GPU. I would need double/triple that amount of GPU cores in my system = 2x76 or more would be great

Out of curiosity, what is your workflow? For what are you using the machine? What is the configuration of your 2019 Mac Pro? Since you did performance comparisons between it and various of the new Apple Silicon M2 systems, how about posting your results? It would be especially interesting, as it seem you are saying that it is faster than an M2 Ultra in your use. Understanding what your use is would help others understand how comparable it would be to their potential use.