Who wants to run the Logic tracks benchmark on their M2/M3?

[eicca]

Logic Pro M1 (Pro & Max) Benchmarks - Music-Prod.com

Here is the Logic Pro X Benchmark project file to use. Just download it, run it and see how many tracks you can get to play.

music-prod.com

Anyone want to give this a whirl and post the score?

I'm very curious since the base M1 scores about the same as a fully-upgraded classic Mac Pro with dual 6-core Xeons. My own Mac Pro isn't even that highly spec'd and I have never pushed its limit. Crazy that my little entry level laptop I'm typing on would decimate my trusty old tower.

 

[ori69]

MaxTech made a comparison of M1 vs M2

 

[Pugly]

I get about 80 tracks on an M1. The M1 is well past the point of having to worry about performance as far as I'm concerned.

I tested how many simple software instrument tracks I could run, with a sample instrument and a few other plugins each. The M1 handled 800 tracks...

 

[ori69]

I did Logic Benchmark today on an Intel i5 6-core 2.5GHz and got 130 tracks.

 

[ori69]

I found a comment on the Logic Benchmark website about MB Air M2:

 

[ori69]

Logic Benchmark Desktop Mac.

 

[eicca]

I’ve seen in multiple places that the M2 consistently scores about 20-25 tracks less than the M1. How is that possible?

That’s even true in the Mini, which isn’t thermally constrained like the Airs.

The heck?

 

[Rnd-chars]

Interesting. I’m not a Logic user so I can’t speak from personal experience, but it looks like a lot more comes into play than CPU alone.

From what I can tell, CPU, RAM, and disk speed can each be a bottleneck leading to System Overload. If that’s true and without knowing the specs of the machines posted, I’d wager the M2 scores lower for a similarly same spec-ed M1 because it uses less NAND for its storage, reducing its throughput and effectively starving Logic for data.

However, I bet if you compared a max spec-ed M1 vs a max spec-ed M2 (which will have identical SSD throughput since the NAND differences only impact the smaller drives), then the M2 would perform faster.

 

[eicca]

Oh yeah forgot about those smaller drives.

Anyone got a 512GB or larger M2 Air/Mini that would be willing to run this benchmark?

 

[ori69]

Oh yeah forgot about those smaller drives.

Anyone got a 512GB or larger M2 Air/Mini that would be willing to run this benchmark?

RAM size and SSD speed don't matter. All that matters in this test is CPU power.
I personally did these tests on a Mac Mini with M2 SSD 256GB, 8GB RAM and SSD 1TB 16GB RAM. And there was no difference in the results.

 

[eicca]

RAM size and SSD speed don't matter. All that matters in this test is CPU power.
I personally did these tests on a Mac Mini with M2 SSD 256GB, 8GB RAM and SSD 1TB 16GB RAM. And there was no difference in the results.

Very strange. There’s no logical (pun intended) reason for the M2 to score 25% lower in that test when it’s more powerful in every conceivable way than M1.

 

[jabbr]

I’m pretty sure the reason is the M1 cores are more stable. Which is significant when it comes to music production, where signal chains mostly get executed on single cores.

M2 has faster single-core performance but is more prone to system overloads because of worse stability. These overloads cause audible pops/clicks and can halt the DAW.
I guess this is mostly a minor trade off. Tracks would bounce faster on the M2

 

[daavee80]

I’m pretty sure the reason is the M1 cores are more stable. Which is significant when it comes to music production, where signal chains mostly get executed on single cores.

M2 has faster single-core performance but is more prone to system overloads because of worse stability. These overloads cause audible pops/clicks and can halt the DAW.
I guess this is mostly a minor trade off. Tracks would bounce faster on the M2

I’m really puzzled/intrigued to know what you mean by “stability” and “system overloads” in this context? If you mean clock rate fluctuation the fan cooled Apple Silicon systems don’t reduce clock rate when saturated.

 

[ori69]

I’m really puzzled/intrigued to know what you mean by “stability” and “system overloads” in this context? If you mean clock rate fluctuation the fan cooled Apple Silicon systems don’t reduce clock rate when saturated.

You're wrong. M2 processors, are the same technology as M1, just slightly overclocked. That's why they heat up faster and get into throttling faster and lower the clock rate faster.
Logic Benchmark test, just like Cinebench loads the cores to the limit. On my computer I tested for half an hour and the whole time the occupation was almost at 100%.

 

[daavee80]

You're wrong. M2 processors, are the same technology as M1, just slightly overclocked. That's why they heat up faster and get into throttling faster and lower the clock rate faster.
Logic Benchmark test, just like Cinebench loads the cores to the limit. On my computer I tested for half an hour and the whole time the occupation was almost at 100%.

View attachment 2227814

But like I said the fan cooled Apple Silicon Macs don’t throttle…

Edit: Ok it looks like certain M2 laptops can throttle under load. Still doesn’t explain the Mac Studio results posted above.

 

[jabbr]

imagine two runners racing, one faster than the other. The faster one is more prone to falling(system overload) but still wins the race(render/bounce). Falling doesn't matter for rendering stuff, but it does for stability of live data streams like tracks in a daw.

In my case where I like being able to monitor at low audio buffers with many effects on a single channel, the M1 Pro is better.

 

[streetfunk]

In my case

Thats the point here. The sole one i´d claim

the M1 Pro is better.

Have you had a M1 beside a M2 ? ( i did. M1 vs. M2pro / same specs otherwise )
If thats true what you are saying, M1 *beside* M2. Then there´s something wrong with your M2.
But if you do just theorize.......?

not shure if the M1pro has double the databus speed vs. a M2.
M2 vs. M2pro, the pro has........(see my story below)
and not sure what macs you compare

The faster one is more prone to falling

Wrong generalisation.
One that runs on the verge of its capacity "might" be more prone for failure.
Do you have evidence that the M2 is running on the verge of the capabilitys of its tech ?
I have a M2pro, and i think its not.

small example vs. "you" having a problem:
i´ve had an M1. At some point in time my Audio system had developed a problem.
I had to switch to another M1, cause i needed a bigger SSD.
I migrated the system.
My problem was no longer present.


my main Audio app runs explicitly on just one core.
But i totally regularly eat that one core. Peanuts with what i´m doing.
I Patch "quite advanced" realtime play patches / realtime-play sounddesign.
Its equivalent to one track in a DAW, musicality wise.
Just alone the higher Databus speed of the M2pro vs. a M1 or M2 (nonpro) makes a big difference vs. how good the system is running.

The Logic test, here in question, is btw. rubbish in my opinion.
Applesilicon macs run best at a buffer setting of 64samples at not too high load, and 128 samples even at higher load.
Also, is the load to the system plugin dependend.
i for example could run a M1 into its knees just alone by the databus speed load.
This, with abvove mentioned patching, which reflects just "one instrument/one track" (running on just one core).......that happened, bevore my CPU ( just one core) was anything close to its limit.

- - - - Databus speed Ladys and Gentleman - - - - -


This logic test should definitly be at a buffer setting of 128 samples.
just my opinion ofcourse. As someone who is not exactly into working with DAWs. and when i do do i have 1-2-3-4 tracks with plugins. In cubase, i was then the other day at above 50% CPU load ( M2pro, 12core).

No generlisations possible !
So much things are totally usecase dependend

 

[jabbr]

The Logic test, here in question, is btw. rubbish in my opinion.
Applesilicon macs run best at a buffer setting of 64samples at not too high load, and 128 samples even at higher load.

I'm not sure what you are trying to say here. Do we know what buffer maxTech used? I would assume whatever the default option is (256 samples I believe).

If anything testing the machines at a lower buffer would only support my point. It takes stable cores to run at low buffers. Since the M2 Pro overloads at a lower track count then we know it's not as stable as the M1 Pro when pushed. This makes sense because the M2 cores are just overclocked M1 cores. Higher throughput at the cost of reduced stability. Gamers overclocking GPUs to the point of crashing their games are very aware of this tradeoff.

I'd rather have the core stability during production over whatever minuscule time I'd save bouncing a track out.

 

[eicca]

Well I did confirm that my dual-X5677 5,1 will handle around 70 tracks. So the base M2 does beat my current setup and would be sufficient for my needs.

 

[ori69]

Logic Benchmark iMac M3, Macbook Pro 16" M3 Pro 12 core and Macbook Pro 14" M3 Pro 10 core

 

[ggCloud]

Logic Benchmark iMac M3, Macbook Pro 16" M3 Pro 12 core and Macbook Pro 14" M3 Pro 10 core

View attachment 2309042

View attachment 2309040

View attachment 2309038

View attachment 2309037

I get the M3 Pro being slower due to fewer P cores but how is the M3 slower than M2 when M3 got improvements all over.

The iMac must have poor cooling compared to the Mac mini. Throttling could explain it.

 

[eicca]

Why TF is the base chip doing LESS tracks with every release…

 

[ggCloud]

Why TF is the base chip doing LESS tracks with every release…

I know that's so weird. The base M3 got upgraded in every possible way. Unlike the M3 Pro the cores stayed the same.

If you see Lukes real world tests M3 is much faster than M2.

So what is up with Logic?

 

[ggCloud]

Looking at video above, I don't think Apple is making Logic a priority.

M3 is better in every test than M2 but Logic. WTF?

 

[eicca]

Looking at video above, I don't think Apple is making Logic a priority.

M3 is better in every test than M2 but Logic. WTF?

Figures that Apple would prioritize developing everything except for ALL THE THINGS I actually use them for.