Apple introduces M2 Ultra

[PAK]

Yes, Cinema 4D (Cinebench) is a better DAW performance benchmark than purely synthetic benchmarks.

D'oh! Better than useless is still useless! Cinebench simply will not inform someone on how specific plugins will perform under specific host software at specific latency settings. It just won't.

Whilst Geekbench does try a broader "synthetic" approach, someone would likewise be completely nuts to buy anything based purely on what its scores say.

People should not base ANY decisions on anything other than the software they will use. Unfortunately, due to the number of variables, this does make things more complicated than is typical for other use cases (And also why many DPC issues get through on PC drivers - because it doesn't matter in many cases, so the issues go noticed until a pesky audio user complains.. )

[PAK]

Well, that's asweome, but, I rather rely on benchmarks which cover many use cases, not just virtual instruments

Ok.. It really only makes sense to care about the software you'll use. I still sometimes use Kontakt. My buffer is usually 128. So it's relevant to me. The issue with many benchmarks is they're frequently not relevant to anything other than silly peepee measuring contests. What software runs on should merely be a means to an end.

Ok, valid choice. But at least acknowledge it contradicts what you said somewhat?

It doesn't as it's a single case scenario, with a single plugin. I have no idea whether or not Kontakt is optimized to work better with Apple's ARM CPU's.

Err.. this pretty much applies to all software.. including benchmarks. They are fairly meaningless if they don't represent a persons actual use case - And benchmarks tend to ignore audio partly because there are so many variables. EG: Imagine a Diva benchmark. Entirely useless if you don't know the patch used as different resonance settings will change the CPU use etc.

Read Jim Roseberry's post above, it explains why it's nonsense to just take a single plugin and test it with two CPU's. Well, maybe not nonsense if you exclusively want to use that plugin, and it's all you will ever use. But, nonsense in terms of determining the performance of a CPU in comparison to others.

What those single results expose is that Cinebench doesn't provide a meaningful metric for audio use because it's not testing those apps, and the apps it is testing tend to work differently. Kontakt is merely one prominent app which demonstrates the picture is more complex than some might like to paint.

[Jac459]

Yes, Cinema 4D (Cinebench) is a better DAW performance benchmark than purely synthetic benchmarks.

Would you mind sharing the link to the benchmark you are talking about? I can't seems to find any that is not with a big UNVERIFIED logo.

[Uncle E]

This thread be like:

-> Apple has a new processor
-> Their new AR headset is shite
-> Apple is a cult
-> Their new CPU won't be as fast as Intel/AMD
-> Their new CPU won't be as fast as Intel/AMD
-> Their new CPU won't be as fast as Intel/AMD
-> Their new CPU won't be as fast as Intel/AMD
-> You can't really make music on a Mac, they're too slow

“My processor has more power than I’ll ever need!”
“My processor has EVEN MORE power than I’ll ever need!”

[Etienne1973]

Jim Roseberry is an expert. That's a fact. He knows what to expect in regard verified benchmark 'Cinebench R23 Multi-Core' results.

Plus here (quote below) is an argument which speaks against that the Apple M2 Ultra can outperform an Intel 13900k.

Of course we don't know yet in the absence of any verified tests. I'd say it's therefore unlikely to do so, if speed has only improved by Apple's own blurb that it's 20% faster then the M1 Ultra. This doesn't bode well for besting the Intel chip based on current results;

https://www.cpu-monkey.com/en/compare_c ... _i9_13900k

Passmark is similar. The M1Ultra is around 40000 for CPU mark, the i9 13900k around 60000;

https://www.cpubenchmark.net/desktop.html

Of course a verified reflection will need to wait, but if I were a betting man.....

[Jac459]

I don't think there is much chance the m2 ultra can compete with the 13900k, one consume 90w, the other 253w...
The power (in)efficiency of the Intel is quite baffling though. I wonder if AMD is much better at this...

[Etienne1973]

Yeah. But AFAIK with proper cooling PC power users got ways around the issue. Sure the power consumption stays.

Edit: I wonder how loud these Intel/AMD workstations are ?

[Jim Roseberry]

The 7950x (like the 13900k) puts off a lot of heat.
So no, AMD isn't better power-consumption wise.

[Jim Roseberry]

As far as benchmarks:
When it comes to PC hardware, I don't look to the Internet, I build/configure and test for maximum performance.
I've been building professionally for 30 years.

I've got a M1 Pro.
I've tested it as well.
Based on that, benchmarks from the M1 Ultra, and leaked benchmarks of the M2... one can fairly accurately project where the M2 falls performance-wise.

Based on Cinema 4D single-core and multi-core benchmark scores, one can fairly accurately get a sense of a CPU's DAW performance.

Multi-Core Cinebench scores:
13900k = 40k
7950x = 38k
12900ks = 28k
M1 Pro = 12k

Ironically, that exactly how those CPUs shake out DAW performance wise.

Using the above scores, let's say the M2 literally doubles the M1's performance.
That puts the M2 Max right about the speed of a 12900k.
Good performance for what it is... but not equal or besting the 13900k.
Again, M2 was designed for small/tight enclosures... not flat out ultimate performance (which would require cooling beyond small-form factor).

I'll grab a M2 Pro and benchmark it.

[Jim Roseberry]

I don't think there is much chance the m2 ultra can compete with the 13900k, one consume 90w, the other 253w...
The power (in)efficiency of the Intel is quite baffling though. I wonder if AMD is much better at this...

You are absolutely correct.
Laws of Thermodynamics.
You have two extremes of the spectrum.

[bmanic]

The low power consumption is exactly what makes Mac laptops so desirable. You get a lot of CPU cycles for the power used, thus the battery life is quite a bit better on Mac laptops, even under heavy loads. They also don't throttle almost at all, even during peak sustained performance so you can relatively safely have a live gig where the computer is going full tilt for 3+ hours. Trying to find a windows laptop that can compete in a similar scenario is extremely difficult (though AMD is close with their latest 7000 series).

If you are building a desktop, especially for studio use where your computer is in a separate room, and you want ultimate CPU performance (and don't care about electricity usage), then intel is the current way to go. Period. There's no way around that fact.

Same for graphics. If you need the absolute power, you go with Intel+Nvidia and eat the electricity bills.

[zvenx]

As far as benchmarks:
When it comes to PC hardware, I don't look to the Internet, I build/configure and test for maximum performance.
I've been building professionally for 30 years.

I've got a M1 Pro.
I've tested it as well.
Based on that, benchmarks from the M1 Ultra, and leaked benchmarks of the M2... one can fairly accurately project where the M2 falls performance-wise.

Based on Cinema 4D single-core and multi-core benchmark scores, one can fairly accurately get a sense of a CPU's DAW performance.

Multi-Core Cinebench scores:
13900k = 40k
7950x = 38k
12900ks = 28k
M1 Pro = 12k

Ironically, that exactly how those CPUs shake out DAW performance wise.

Using the above scores, let's say the M2 literally doubles the M1's performance.
That puts the M2 Max right about the speed of a 12900k.
Good performance for what it is... but not equal or besting the 13900k.
Again, M2 was designed for small/tight enclosures... not flat out ultimate performance (which would require cooling beyond small-form factor).

I'll grab a M2 Pro and benchmark it.

Hi, I am curious if you have benchmarks for the 13900k with track counts etc that you can share (like https://dawbench.com/)

thanks
rsp

[Jim Roseberry]

The low power consumption is exactly what makes Mac laptops so desirable. You get a lot of CPU cycles for the power used, thus the battery life is quite a bit better on Mac laptops, even under heavy loads. They also don't throttle almost at all, even during peak sustained performance so you can relatively safely have a live gig where the computer is going full tilt for 3+ hours. Trying to find a windows laptop that can compete in a similar scenario is extremely difficult (though AMD is close with their latest 7000 series).

If you are building a desktop, especially for studio use where your computer is in a separate room, and you want ultimate CPU performance (and don't care about electricity usage), then intel is the current way to go. Period. There's no way around that fact.

Same for graphics. If you need the absolute power, you go with Intel+Nvidia and eat the electricity bills.

Well said.

FWIW, If you get the right cooler (large/expensive), the 13900k can run quiet.
I've got one sitting in the lower right side of my studio desk.

Any type of fan/computer noise drives my crazy.
I've returned a Neve summing mixing... and got rid of the Axe-FX II and III because of this.

Due to the lower power consumption, the M1 and M2 machines do run quiet.

[revvy]

This thread be like:

-> Apple has a new processor
-> Their new AR headset is shite
-> Apple is a cult
-> Their new CPU won't be as fast as Intel/AMD
-> Their new CPU won't be as fast as Intel/AMD
-> Their new CPU won't be as fast as Intel/AMD
-> Their new CPU won't be as fast as Intel/AMD
-> You can't really make music on a Mac, they're too slow

“My processor has more power than I’ll ever need!”
“My processor has EVEN MORE power than I’ll ever need!”

“Mac users are even more freakish than Linux users (except for you and my mother)”
“Apple make nice phones, very quiet”

[Uncle E]

(except for you and my mother)

Where’s that throw up emoji?