Apple announces new Mac Mini, Air + 13" MBP featuring their own M1 chip.

[KenjiDeVries]

It was the latest Logic update available and tested with Logics factory plugins (one session only fx, one session mainly synths) and with native 3rd party plugins as well as Rosetta 3rd party plugins. Rosetta vs. native wasn't that much of a different, about 10-15% on averge. I just saw Tracktion seems to be available as native ARM version too, so I'll give this one a try too. I felt either like the M1 was in some power saving mode or switched to it after some time or didn't fully utilize its cores like expected (even after explicitely telling it via Logic). If this is true and I can't change that then I'd disqualify it as a daily workhorse machine.

[frizzbee]

I noticed Logic didn't have all cores enabled for processing. I enabled them but it didn't change anything.

german youtuber had tested mac mini as well and noticed it too. he got flamed by apple fanboys because "he must have done something wrong" and we know it's never apples fault now there are more people coming forward, interesting...

If this is true and I can't change that then I'd disqualify it as a daily workhorse machine.

imo it's too soon anyway to tell anything about it. first genration of chip that has to prove its power on desktops, not much native software available besides the non saying standard benchmarks. if it feels like a mobile chip in a bigger case then i'd agree.

[Sascha Franck]

Had the chance to test the M1 (mac mini) over the weekend. The performance was not bad but somehow not that impressive. During tests with Logic, StudioOne and ProTools (which all had similar results) I noticed Logic didn't have all cores enabled for processing. I enabled them but it didn't change anything.

Quite the opposite of what I've seen in other reports. Weird.

[Sascha Franck]

german youtuber had tested mac mini as well and noticed it too.

Talking about Musotalk?

[Sascha Franck]

Btw, in case you're using Logic's internal plugins, it seems as if Alchemy isn't fully silicon-adapted yet (resulting in somewhat worse performance than on Intel machines as it seems), so in case you were using it, it might blurry the picture.

[Markus Krause]

The M1 has 4 fast cores (firestorm) and 4 additional stripped-down slow cores (icestorm). The fast cores are around 4 times faster than the slow ones. It is not a real 8 core machine. It is more similar to a quadcore CPU with hyperthreading or something like a 5 core machine. I assume the task manager distributes processes with high priority like audio to the 4 fast cores. Processes with low priority seem to go to the slow ones.

[Sascha Franck]

Well, by now there's plenty of Logic tests with the 8 cores being used almost equally. And some A/B before/after tests also showing a significant difference.

[KenjiDeVries]

Btw, in case you're using Logic's internal plugins, it seems as if Alchemy isn't fully silicon-adapted yet (resulting in somewhat worse performance than on Intel machines as it seems), so in case you were using it, it might blurry the picture.

Yes indeed I noticed something like this with the Logic factory synth test. But even there the 4 more cores didn't change anything.

The M1 has 4 fast cores (firestorm) and 4 additional stripped-down slow cores (icestorm). The fast cores are around 4 times faster than the slow ones. It is not a real 8 core machine. It is more similar to a quadcore CPU with hyperthreading or something like a 5 core machine. I assume the task manager distributes processes with high priority like audio to the 4 fast cores. Processes with low priority seem to go to the slow ones.

Okay thank you, that indeed would be an explanation. I just read that the M1 seems to be able to boost the performance for 15 to 20 minutes before cooling down again. Maybe that's the weird performance drop I noticed.

[Ploki]

I mean, it performs better than any 4-core machine, any 6-core machine, and any mobile 8-core machine, and it's for all intents and purposes and entry level 4-core

Why are people disappointed?

[KenjiDeVries]

Well right now with the Mac Mini I'm not much better than with my girlfriends 2017 Mac Book (either MacOS or Bootcamp Windows). And I think that's not supposed to be like that And I don't see many options to change anything for better performance. I also do not want to spend time digging what's wrong. I mean it's strange enough Logic isn't setup for max performance by default.

[Sascha Franck]

Yes indeed I noticed something like this with the Logic factory synth test. But even there the 4 more cores didn't change anything.

First off: As said, some people seem to notice a very clear (almost drastic) improvement when activating all cores in Logic. You might want to check this thread at GS:
https://www.gearslutz.com/board/music-c ... audio.html
But there's also one video where the guy switches to something else (I think it's just the macOS activity monitor) after enabling all cores in Logic, followed by an overload.
My assumption would be that with 4 cores enabled, the high performance ones are sort of reserved for Logic (at least as long as its playing) and you could likely do anything else not using up much power in parallel and these things would then be taken care of by the low performance cores. But once you activate all cores in Logic, the possible CPU sharing might then throw Logic off a bit.
Could perhaps be tested by running nothing else but Logic and then comparing performance while switching between apps. One time with just 4 cores, one time with all cores enabled within Logic.

[Ploki]

Well right now with the Mac Mini I'm not much better than with my girlfriends 2017 Mac Book (either MacOS or Bootcamp Windows). And I think that's not supposed to be like that And I don't see many options to change anything for better performance. I also do not want to spend time digging what's wrong. I mean it's strange enough Logic isn't setup for max performance by default.

it's not?
Every benchmark i saw in logic native performs better than a 2020 i9 8-core 16".
It wipes the floor with my i7 Mini.

There's no 2017 MacBook that's better or even near it. Even the 15" is worse than my 2018 13" i5.

[Tilly]

Just saying that benchmarks often are a bit different from daily work and long time performance need. Had this issue recently with a AMD vs Intel CPU where the AMD outperformed Intel but only was able to work in spikes before it dropped while after all the Intel performance was steady.

[Markus Krause]

Running a plugin on a fast core does result in a lower possible latency than running it on a slow core. I think this is the reason why Logic uses only the 4 fast cores by default. It is avoiding crackles.
A signal chain that runs in real-time on several processes is only as strong as its weakest point

[Sascha Franck]

Running a plugin on a fast core does result in a lower possible latency than running it on a slow core. I think this is the reason why Logic uses only the 4 fast cores by default. It is avoiding crackles.
A signal chain that runs in real-time on several processes is only as strong as its weakest point

In Logic, all playback-only tracks are set to a buffer size of 1024 automatically. That's why there's no issues activating all cores.