A technical support community for Apple Logic Pro users.

 
joshj
Topic Author
Posts: 422
Joined: Wed Dec 14, 2011 12:57 pm

MacMini M1 Logic Benchmark results (strange)

Thu Nov 19, 2020 12:24 pm

I just ran these 2 Logic benchmarks on my 2018 MacMini 6-core i7 32GB RAM against the new M1 Mac Mini 8-core 8GB RAM. I got the following results;

Sculpture Benchmark (Per channel: Sculpture, Channel EQ, Multipressor, Chorus, AutoFilter, PlatinumVerb):
MacMini M1 8-core 8GB RAM: 109 tracks (before System Overload)
MacMini i7 6-core 32GB RAM: 82 tracks

Alchemy Benchmark (Per channel: Alchemy only):
MacMini M1 8-core 8GB RAM: 126 tracks
MacMini i7 6-core 32GB RAM: 146 tracks

I have some questions;
Why does my machine (i7) perform better with the supposedly more system intensive Sculpture benchmark and the M1 Mini (more cores) perform better with the Alchemy Benchmark? I would have expected the M1 to outperform on both. How much is the difference in RAM factoring into these results?
Also, could anyone hedge a guess at how many more tracks the M1 would likely be able to handle if it was the 16GB RAM version?

Here's the benchmark if anyones interested. The creator has posted on LPH before: https://music-prod.com/logic-pro-x-benchmarks/
Logic Pro 10.5 | OSX 10.14
MacMini 2018 3.2GHz 6 Core i7, 30GB RAM
Focusrite Scarlett 2i2
 
User avatar
des99
Posts: 5479
Joined: Tue Jan 27, 2015 12:08 pm

Re: MacMini M1 Logic Benchmark results (strange)

Thu Nov 19, 2020 12:52 pm

It might be that Alchemy has a more complex audio engine that doesn't optimize for the M1 as well (or perhaps they are still working on optimising it because it was very heavily optimised for Intel).

RAM is unlikely to make much of a practical difference, as long as you weren't maxing it out to a lot of swap (which you can check with Activity Monitor). More RAM does not make your CPU faster, it just reduces bottlenecks elsewhere in the system *if* a lack of RAM was slowing things down.

What patch was playing in Alchemy? A sample based patch? A VA-based patched? An additive/granular patch? Does it make a difference in your benchmarks? (as they are very different synthesis modes with different performance characteristics.)
mu:zines | music magazine archive | difficultAudio
 
TheRdungeon
Posts: 315
Joined: Sat May 14, 2011 7:41 pm

Re: MacMini M1 Logic Benchmark results (strange)

Fri Nov 20, 2020 2:51 am

I would guess that only the main Logic app itself is optimised for apple silicon, and maybe a few plugins. Alchemy is essentially a third party plugin they purchased and put in Logic, so it could still be running in Rosetta mode at this point
2.3GHz Quad MBP - Mid '12 15"
Apogee Duet 2
Logic 10.4.7
OS X 10.14.6
 
User avatar
Ploki
Posts: 4529
Joined: Tue Mar 04, 2008 2:59 am
Location: Slovenia
Contact:

Re: MacMini M1 Logic Benchmark results (strange)

Fri Nov 20, 2020 3:28 am

TheRdungeon wrote:
I would guess that only the main Logic app itself is optimised for apple silicon, and maybe a few plugins. Alchemy is essentially a third party plugin they purchased and put in Logic, so it could still be running in Rosetta mode at this point

yep.
and the % difference seems to be pretty inline with other rosetta v native
<desktop pending> | 13" M1 Pro | Big Sur 11.1 | Logic 10.6.1
RME FireFace 800 / UFX+ | ROLI Seaboard RISE 25 | ROLI Blocks | nOb Control | StreamDeck
 
joshj
Topic Author
Posts: 422
Joined: Wed Dec 14, 2011 12:57 pm

Re: MacMini M1 Logic Benchmark results (strange)

Fri Nov 20, 2020 4:56 am

des99 wrote:
What patch was playing in Alchemy


I think it’s just a basic synth preset
Logic Pro 10.5 | OSX 10.14
MacMini 2018 3.2GHz 6 Core i7, 30GB RAM
Focusrite Scarlett 2i2
 
joshj
Topic Author
Posts: 422
Joined: Wed Dec 14, 2011 12:57 pm

Re: MacMini M1 Logic Benchmark results (strange)

Fri Nov 20, 2020 5:07 am

TheRdungeon wrote:
I would guess that only the main Logic app itself is optimised for apple silicon, and maybe a few plugins. Alchemy is essentially a third party plugin they purchased and put in Logic, so it could still be running in Rosetta mode at this point


Got you. So I should base my judgment of the M1 on the Sulpture Benchmark (roughly 30tracks more able than the i7). And that performance boost will translate over the next few months to 3rd party plugins? So plugin developers have to rewrite for just BigSur or for how their plugins run on M1 specifically? Is that a big change in the code?
Logic Pro 10.5 | OSX 10.14
MacMini 2018 3.2GHz 6 Core i7, 30GB RAM
Focusrite Scarlett 2i2
 
TheRdungeon
Posts: 315
Joined: Sat May 14, 2011 7:41 pm

Re: MacMini M1 Logic Benchmark results (strange)

Fri Nov 20, 2020 5:17 am

joshj wrote:
TheRdungeon wrote:
I would guess that only the main Logic app itself is optimised for apple silicon, and maybe a few plugins. Alchemy is essentially a third party plugin they purchased and put in Logic, so it could still be running in Rosetta mode at this point


Got you. So I should base my judgment of the M1 on the Sulpture Benchmark (roughly 30tracks more able than the i7). And that performance boost will translate over the next few months to 3rd party plugins? So plugin developers have to rewrite for just BigSur or for how their plugins run on M1 specifically? Is that a big change in the code?


They have to re-write for Apple Silcon/M1 specifically. For some developers this could take a while, for example I emailed TB Pro Audio because DSEQ is a huge part of my workflow and they said probably around 3 months. But they're a small developer, so who knows how long the bigger guys will take
2.3GHz Quad MBP - Mid '12 15"
Apogee Duet 2
Logic 10.4.7
OS X 10.14.6
 
User avatar
des99
Posts: 5479
Joined: Tue Jan 27, 2015 12:08 pm

Re: MacMini M1 Logic Benchmark results (strange)

Fri Nov 20, 2020 5:19 am

They have to recompile for M1, and in some cases, where the developer has done optimisations for specific features of the Intel architecture for performance reasons, they may have to rework some of those parts of the code.

All this stuff will shake out over the coming year (welcome to transitions!) but in general, you can see how much more powerful and power efficient the M1 is over the Intel chips, and that will have a huge positive affect on running M1-based Mac systems in general.
mu:zines | music magazine archive | difficultAudio
 
User avatar
skijumptoes
Posts: 1970
Joined: Wed Jul 15, 2009 3:09 pm

Re: MacMini M1 Logic Benchmark results (strange)

Fri Nov 20, 2020 6:07 am

Curious how the AVX2 reliant plugins port across. One presumes that's a more considerable hurdle to leap.

Also, wonder how something like Massive X that does require it performs in Rosetta2 at this initial stage.
 
joshj
Topic Author
Posts: 422
Joined: Wed Dec 14, 2011 12:57 pm

Re: MacMini M1 Logic Benchmark results (strange)

Fri Nov 20, 2020 6:54 am

Ploki wrote:
TheRdungeon wrote:
I would guess that only the main Logic app itself is optimised for apple silicon, and maybe a few plugins. Alchemy is essentially a third party plugin they purchased and put in Logic, so it could still be running in Rosetta mode at this point

yep.
and the % difference seems to be pretty inline with other rosetta v native


Are talking about Rosetta apps or plugins? What % lower performance difference at people seeing on the whole with Rosetta?
Logic Pro 10.5 | OSX 10.14
MacMini 2018 3.2GHz 6 Core i7, 30GB RAM
Focusrite Scarlett 2i2
 
User avatar
Ploki
Posts: 4529
Joined: Tue Mar 04, 2008 2:59 am
Location: Slovenia
Contact:

Re: MacMini M1 Logic Benchmark results (strange)

Fri Nov 20, 2020 7:06 am

joshj wrote:
Ploki wrote:
yep.
and the % difference seems to be pretty inline with other rosetta v native


Are talking about Rosetta apps or plugins? What % lower performance difference at people seeing on the whole with Rosetta?

20-30%
<desktop pending> | 13" M1 Pro | Big Sur 11.1 | Logic 10.6.1
RME FireFace 800 / UFX+ | ROLI Seaboard RISE 25 | ROLI Blocks | nOb Control | StreamDeck
 
User avatar
skijumptoes
Posts: 1970
Joined: Wed Jul 15, 2009 3:09 pm

Re: MacMini M1 Logic Benchmark results (strange)

Fri Nov 20, 2020 8:26 am

Ploki wrote:
joshj wrote:
Are talking about Rosetta apps or plugins? What % lower performance difference at people seeing on the whole with Rosetta?

20-30%

20-30% of what though? Of what the M1 'could' do natively? Or what their previous machine could do?

I don't know how the comparison is formed, Because a 20-30% loss, as long as you're on a machine that is 30% faster than your old is still a gain even on Rosetta 2, right?
 
User avatar
Ploki
Posts: 4529
Joined: Tue Mar 04, 2008 2:59 am
Location: Slovenia
Contact:

Re: MacMini M1 Logic Benchmark results (strange)

Fri Nov 20, 2020 9:38 am

skijumptoes wrote:
Ploki wrote:
20-30%

20-30% of what though? Of what the M1 'could' do natively? Or what their previous machine could do?

I don't know how the comparison is formed, Because a 20-30% loss, as long as you're on a machine that is 30% faster than your old is still a gain even on Rosetta 2, right?

20-30% compared to same process running natively :)

and yes, you're right. M1 13" via rosetta 2 is faster than any intel 13".
<desktop pending> | 13" M1 Pro | Big Sur 11.1 | Logic 10.6.1
RME FireFace 800 / UFX+ | ROLI Seaboard RISE 25 | ROLI Blocks | nOb Control | StreamDeck
 
User avatar
Dewdman42
Posts: 3304
Joined: Tue Sep 09, 2014 3:01 pm
Location: Salt Lake City, UT

Re: MacMini M1 Logic Benchmark results (strange)

Fri Nov 20, 2020 11:57 am

TheRdungeon wrote:
I would guess that only the main Logic app itself is optimised for apple silicon, and maybe a few plugins. Alchemy is essentially a third party plugin they purchased and put in Logic, so it could still be running in Rosetta mode at this point


The built in plugins are not actually plugins they are built in to the main executable. So if logicpro is running the Arm binary then so are they.

Differences in performance are more likely related to either specific intel optimizations such as AVX which are not in ARM, or perhaps related to other ssd streaming bottlenecks.
OSX 10.15 (Catalina) on OpenCore - Logic Pro 10.6.1, VePro7, Mainstage3
5,1 MacPro 3.46ghz x 12 96gb ram
 
User avatar
Glenn Lancaster
Posts: 1189
Joined: Fri Jun 20, 2008 2:06 pm
Location: The Windmills of my Mind

Re: MacMini M1 Logic Benchmark results (strange)

Fri Nov 20, 2020 11:49 pm

Looks like my next 'puter will be interesting, provided this one holds on a couple of more years. Or should I wait for their iQuantum chip?
2020 M1 Mini, macOS 11.4, Logic 10.6.3, RME FF802, Novation 61SL Mk II. Martins and Fenders and Gibsons, oh my!
 
joshj
Topic Author
Posts: 422
Joined: Wed Dec 14, 2011 12:57 pm

Re: MacMini M1 Logic Benchmark results (strange)

Sun Nov 22, 2020 4:15 am

TheRdungeon wrote:
joshj wrote:

Got you. So I should base my judgment of the M1 on the Sulpture Benchmark (roughly 30tracks more able than the i7). And that performance boost will translate over the next few months to 3rd party plugins? So plugin developers have to rewrite for just BigSur or for how their plugins run on M1 specifically? Is that a big change in the code?


They have to re-write for Apple Silcon/M1 specifically. For some developers this could take a while, for example I emailed TB Pro Audio because DSEQ is a huge part of my workflow and they said probably around 3 months. But they're a small developer, so who knows how long the bigger guys will take



You think the big guys will take longer or shorter than the small developers?
Logic Pro 10.5 | OSX 10.14
MacMini 2018 3.2GHz 6 Core i7, 30GB RAM
Focusrite Scarlett 2i2
 
User avatar
des99
Posts: 5479
Joined: Tue Jan 27, 2015 12:08 pm

Re: MacMini M1 Logic Benchmark results (strange)

Sun Nov 22, 2020 4:41 am

Predicting how long software will take is notoriously problematic.

Personally I don't really care too much, because it's the sensible thing during transitions to keep on working as you are, and casually monitor what's going on from the sidelines. Then once everything's settled down and shaken out, then start thinking about moving over, and what the pain points still are.
mu:zines | music magazine archive | difficultAudio
 
User avatar
Ploki
Posts: 4529
Joined: Tue Mar 04, 2008 2:59 am
Location: Slovenia
Contact:

Re: MacMini M1 Logic Benchmark results (strange)

Sun Nov 22, 2020 4:43 am

joshj wrote:
TheRdungeon wrote:

They have to re-write for Apple Silcon/M1 specifically. For some developers this could take a while, for example I emailed TB Pro Audio because DSEQ is a huge part of my workflow and they said probably around 3 months. But they're a small developer, so who knows how long the bigger guys will take



You think the big guys will take longer or shorter than the small developers?


really doesn't matter how big/small dev is, but how old their framework is and how reliant are they to x86 instruction set.
U-he said it's far less problematic than they anticipated, Audiothing said it's not problematic since they're on Juce/Xcode and just need to recompile for ARM, Tone2 said it's a bit problematic but one of their synths is already ARM ready (Warlock), Reaper is already beta for ARM.

So it really depends on what you depend on for programming your stuff :)
<desktop pending> | 13" M1 Pro | Big Sur 11.1 | Logic 10.6.1
RME FireFace 800 / UFX+ | ROLI Seaboard RISE 25 | ROLI Blocks | nOb Control | StreamDeck
 
joshj
Topic Author
Posts: 422
Joined: Wed Dec 14, 2011 12:57 pm

Re: MacMini M1 Logic Benchmark results (strange)

Sun Nov 22, 2020 5:33 am

Ploki wrote:
joshj wrote:


You think the big guys will take longer or shorter than the small developers?


really doesn't matter how big/small dev is, but how old their framework is and how reliant are they to x86 instruction set.
U-he said it's far less problematic than they anticipated, Audiothing said it's not problematic since they're on Juce/Xcode and just need to recompile for ARM, Tone2 said it's a bit problematic but one of their synths is already ARM ready (Warlock), Reaper is already beta for ARM.

So it really depends on what you depend on for programming your stuff :)



I’m more reliant on my mix chain than synths at the moment. FabFilter, Waves, iZotope and Melodyne. Don’t suppose you known anything about how prepared those developers are?
Logic Pro 10.5 | OSX 10.14
MacMini 2018 3.2GHz 6 Core i7, 30GB RAM
Focusrite Scarlett 2i2
 
User avatar
Ploki
Posts: 4529
Joined: Tue Mar 04, 2008 2:59 am
Location: Slovenia
Contact:

Re: MacMini M1 Logic Benchmark results (strange)

Sun Nov 22, 2020 5:44 am

FabFilter will work via rosetta 2 on Logic, and native version is in the works:
https://www.fabfilter.com/forum/6375/ma ... patibility
iZotope
https://support.izotope.com/hc/en-us/ar ... 11-Big-Sur

what i use from iZotope works on Big Sur - i imagine that they should work via Rosetta2.

Melodyne4 works in Big Sur here normally, 5 i supposed should as well, haven't found anything about Celemony re: M1.

As far as Waves go, that's probably going to be problematic. They're not the fastest to react and they tend to charge for compatibility updates - best bet for now is hope they work via rosetta. I also don't have any partially because of the aforementioned reasons :D
<desktop pending> | 13" M1 Pro | Big Sur 11.1 | Logic 10.6.1
RME FireFace 800 / UFX+ | ROLI Seaboard RISE 25 | ROLI Blocks | nOb Control | StreamDeck