New Logic Multitrack Benchmark Test

changing process buffer range to "medium" from large somehow increased native track count to 37 lol.

also i don't get different results changing buffersize.

in general, i observed M1 behaving ridiculously well at small buffers - even in fully loaded projects. Sometimes i almost complete my mix before even figuring out that "o hey i'm at 64 buffersize"

Ok Pro Tools sucks...

I have, guess what, 34 tracks on a 9900K running 5 GHz on all cores, double the performance in cinebench and faster than the 12 Core nMac Pro 2019!

34 tracks, which means the M1 performs in Logic better than a 12 Core MP 2019 in Pro Tools.

Ok Pro Tools sucks...

 

I have, guess what, 34 tracks on a 9900K running 5 GHz on all cores, double the performance in cinebench and faster than the 12 Core nMac Pro 2019!

 

34 tracks, which means the M1 performs in Logic better than a 12 Core MP 2019 in Pro Tools.

The first time in Logic’s history i feel now the hardware is not holding back my creativity.

This is with the M1 chip. I am only a bedroom musician. I can throw at it a lot of CPU heavy synths, and the GUI is snappy, just does what i tell it to do immediately and well. No workarounds, bouncing etc that just kills my creativity very quickly. And.....OMG the biggest thing for me: does it without a fan. The M1 Air it is. I record with a microphone right beside my computer.

Will be interesting to see the next version of the AIR.

Can Apple up the power further without a fan?

changing process buffer range to "medium" from large somehow increased native track count to 37 lol.

also i don't get different results changing buffersize.

in general, i observed M1 behaving ridiculously well at small buffers - even in fully loaded projects. Sometimes i almost complete my mix before even figuring out that "o hey i'm at 64 buffersize"

Can you take a screenshot of your session playing back showing the track count like me below? And the CPU meter?

So that my disappointing 35 tracks of my 9900KS prowess, running 5 GHz on all cores in Pro Tools Ultimate.

Ok Pro Tools sucks...

 

I have, guess what, 34 tracks on a 9900K running 5 GHz on all cores, double the performance in cinebench and faster than the 12 Core nMac Pro 2019!

 

34 tracks, which means the M1 performs in Logic better than a 12 Core MP 2019 in Pro Tools.

 

 

The first time in Logic’s history i feel now the hardware is not holding back my creativity.

This is with the M1 chip. I am only a bedroom musician. I can throw at it a lot of CPU heavy synths, and the GUI is snappy, just does what i tell it to do immediately and well. No workarounds, bouncing etc that just kills my creativity very quickly. And.....OMG the biggest thing for me: does it without a fan. The M1 Air it is. I record with a microphone right beside my computer.

Will be interesting to see the next version of the AIR.

Can Apple up the power further without a fan?

Can you do the benchmark on your M1 and show me the result?

https://drive.google.com/file/d/1p2P8-U3Ydwn_TxtGlyTEVpbheSCthN4F/view?usp=sharing

maxed out tho.

it's running on a 4K (4096x2304) external screen @hidpi, not sure if it affects it, but i doubt i'd get more tracks running on the 13" screen.

Hey guys,

can someone please explain in detail which effects exactly are on every a single track channel, on sends and on busses or if its just the main mix bus...

I got a little confused which effects were used in the final test, after reading trough the text below:

"Every track contains Fabfilter's Comp, EQ and De-Esser; additionally you can activate bypassed Gate, MB, Saturn and Valhalla Vintage Verb but I mainly kept it to Comp, EQ and De-Esser, which is traditional.

Each channel sends to 8 FX busses with Valhalla VV, Fabfilter Reverb and some logic stock ones.

On the mixbus is Saturn, Comp, MB and Pro L2."

A few posts later I also saw that the logic effects where not used in the final benchmark tests, but on the pictures that were attached I can't really see the routing and effects on the sends, just the effects on a single track in Pro Tools, which I saw were just Fabfilter's Pro Q-3, C-2 and DS.

I'm asking for a detailed effects chain on all the sends and busses, because I can't have a look at the project files since I don't have Logic and would like to benchmark this on another DAW.

Many thanks

Hey guys,

 

can someone please explain in detail which effects exactly are on every a single track channel, on sends and on busses or if its just the main mix bus...

I got a little confused which effects were used in the final test, after reading trough the text below:

"Every track contains Fabfilter's Comp, EQ and De-Esser; additionally you can activate bypassed Gate, MB, Saturn and Valhalla Vintage Verb but I mainly kept it to Comp, EQ and De-Esser, which is traditional.

 

Each channel sends to 8 FX busses with Valhalla VV, Fabfilter Reverb and some logic stock ones.

 

On the mixbus is Saturn, Comp, MB and Pro L2."

 

 

A few posts later I also saw that the logic effects where not used in the final benchmark tests, but on the pictures that were attached I can't really see the routing and effects on the sends, just the effects on a single track in Pro Tools, which I saw were just Fabfilter's Pro Q-3, C-2 and DS.

 

I'm asking for a detailed effects chain on all the sends and busses, because I can't have a look at the project files since I don't have Logic and would like to benchmark this on another DAW.

 

Many thanks

It's Pro Q, Pro C and Pro De-Ess on every track.

Every track also sends to 4 FX busses: 2x Valhalla and 2 x Pro R

All going into Saturn, Pro C, Pro MB and Pro L2.

Thanks for the info.

So if I understood correctly all the previous posts, the maximum number of tracks playing simultaneously with these kind of processing/routing was 35 tracks on M1 Mac and 34 on PC with 9900KS PC? And both running on 1024 buffer length?

I must have missed some other important detail I guess... or you probably tested these two tests on some super low buffer lengths, because on 1024 buffer, I can easily get 40 tracks playing ocean eyes audio file simultaneously with all of these processing/routing and 40 google chrome tabs opened as well and a mid tier cpu, Ryzen 3600 barely hits 40%.

Ok just tested this also on my home pc with almost identical hardware, except for soundcard and gpu and turned down the buffer to 64 and it pushed the cpu usage on average 50% with occasional peak at 60%, so it must be something else than buffer length that I'm missing in this test I guess.

So I made 40 identical audio tracks with the provided "ocean eyes" audio clip, every track has Pro-Q 3, Pro-C 2 and Pro-DS. Each of those 40 audio tracks are routed to 4 FX sends, each send has 2x Pro-R and 2x Valhalla Vintage Verb on it and then all of these 40 tracks go into the final mix channel which has Saturn 2, Pro-C 2, Pro-MB and Pro-L 2 on it. Is this the same routing you guys used? Did you maybe use max Linear phase function on Pro-Q 3?

Ok just tested this also on my home pc with almost identical hardware, except for soundcard and gpu and turned down the buffer to 64 and it pushed the cpu usage on average 50% with occasional peak at 60%, so it must be something else than buffer length that I'm missing in this test I guess.

 

So I made 40 identical audio tracks with the provided "ocean eyes" audio clip, every track has Pro-Q 3, Pro-C 2 and Pro-DS. Each of those 40 audio tracks are routed to 4 FX sends, each send has 2x Pro-R and 2x Valhalla Vintage Verb on it and then all of these 40 tracks go into the final mix channel which has Saturn 2, Pro-C 2, Pro-MB and Pro-L 2 on it. Is this the same routing you guys used? Did you maybe use max Linear phase function on Pro-Q 3?

The pro Q has a high oversampling setting, that's why it only is at these low tracks, you'd have to download the session I uploaded.

I'll do a different test again with everything set to zero latency tomorrow.

Ok cool, that would be great to see the zero latency bench with 1024 and on 64 buffer.

I just tested now all 40 tracks, same routing, but with Pro-Q 3 set to Maximum Linear Phase, so the highest possible setting I know of on this plugin and the cpu usage is around 58%.

Ok cool, that would be great to see the zero latency bench with 1024 and on 64 buffer.

I just tested now all 40 tracks, same routing, but with Pro-Q 3 set to Maximum Linear Phase, so the highest possible setting I know of on this plugin and the cpu usage is around 58%.

Yeah it is partly also due to Pro Tools, which is the most expensive DAW and predominantly used by all the "Pros", performing so badly.

Why don't you just open the session I made?

https://drive.google.com/file/d/1p2P8-U ... sp=sharing

But yeah, maybe the Pro Q 3 in the highest oversampling is causing issues.

I just wanted to do something that scales well for more powerful systems instead of doing 200+ tracks.

Maybe you could set up a session doing the following: Pro Q3, Pro C2, Pro De-esser, Pro MB, Pro-G, Pro L2, Pro Reverb and Valhalla Vintage Verb, all set to zero latency and then duplicate the tracks until you run out of CPU?

I can't do it until Wednesday.

@Ploki maybe you as well?

opened everything you said in the same order, with default presets, disabled oversampling and lookahead, no sends, nothing on the master bus.

85 tracks on my default settings (128 buffersize, medium process buffer range)

105 on max settings 1024 buffer, large process buffer range

gotta note i'm doing this on battery power and on my lap

opened everything you said in the same order, with default presets, disabled oversampling and lookahead, no sends, nothing on the master bus.

85 tracks on my default settings (128 buffersize, medium process buffer range)

 

105 on max settings 1024 buffer, large process buffer range

 

gotta note i'm doing this on battery power and on my lap

Please let me know if I understood you correctly, with the previous processing/routing (Q 3, C 2 and DS on every single track, routed to 4 bus sends with 2x R and 2x Valhalla on each bus and then everything going into master with Saturn, C 2, MB and L 2) you could process maximum of 35 tracks with max buffer?

And with current routing Q 3, C 2, DS, MB, L 2, R and Valhalla on every single track, you could process maximum of 105 tracks with max buffer?

If that is so, then some routing or plugin settings are still not the same in these tests.

Because when I routed the tracks like explained in the first test (with fx sends and plugins on master), I could do max 117 tracks on max buffer, but in the second test (Q 3, C 2, DS, MB, L 2, R and Valhalla on every single track) I could do max 35 tracks on max buffer, which is again a super weird gap between our results of both tests, so yours 35 to mine 117 in the first test and yours 105 to mine 35 in the second test... the reasonable gaps between different results should be like 10-20 tracks the most IMO.

Btw in the first and second test I had all the plugins on default settings, without any oversampling and lookahead, exactly like you did it.

Oh well, it was fun to try some benching anyway and I hope you guys can do some more benching in the future, always love to read this kind of stuff.

opened everything you said in the same order, with default presets, disabled oversampling and lookahead, no sends, nothing on the master bus.

85 tracks on my default settings (128 buffersize, medium process buffer range)

 

105 on max settings 1024 buffer, large process buffer range

 

gotta note i'm doing this on battery power and on my lap

How much storage should I get for my M1?

512 or 1 TB?

opened everything you said in the same order, with default presets, disabled oversampling and lookahead, no sends, nothing on the master bus.

85 tracks on my default settings (128 buffersize, medium process buffer range)

 

105 on max settings 1024 buffer, large process buffer range

 

gotta note i'm doing this on battery power and on my lap

 

How much storage should I get for my M1?

 

512 or 1 TB?

mini or macbook? if it's the mini, it sits on your desk, and you can always archive or store extra work on an external. if it's a macbook... might as well go for 1TB... since you're mobile, and good to have more space...

Alright, so I did a test where Plugins would not introduce Sample latency to keep the test purely CPU, here is my result with all the settings:

I've removed L2 because it doesn't fully run on the CPU and it introduces sample latency and no reverb because it's very dynamic and creates CPU overhead.

What's with the windows/pro tools posts? ...Am i missing something?

Well I'd like to compare it especially with Pro Tools running a powerful desktop processor versus optimized M1 in Logic with the same plugins.

Well I'd like to compare it especially with Pro Tools running a powerful desktop processor versus optimized M1 in Logic with the same plugins.

As this benchmark was specifically designed to utilise Logic's stock plugins, it's confusing coming back and reading all this additional discussion on Pro Tools, Windows and third party plugs. I was thinking there had been a Logic update when i saw those different mixer screenshots lol.

In hindsight would've been nice if this was asked on it's own thread, and you could go into more detail there too. See - Right now, if someone posts their results i'm not sure if the benchmark relates to the original Logic one, or this new fabfilter/valhalla based test.

Plus, one session compared to another which is running on a different plugin architecture, different daw on a different operating system isn't going to be a very reliable comparison. If you got 100 tracks, and someone else had 80 or 120. What does it actually tell you(?) A difference like that could come down to manufacturer drivers, OS System level services, and type of interface the audio device is connected to (PCIe/USB/Thunderbolt etc.).

Well I'd like to compare it especially with Pro Tools running a powerful desktop processor versus optimized M1 in Logic with the same plugins.

As this benchmark was specifically designed to utilise Logic's stock plugins, it's confusing coming back and reading all this additional discussion on Pro Tools, Windows and third party plugs. I was thinking there had been a Logic update when i saw those different mixer screenshots lol.

 

In hindsight would've been nice if this was asked on it's own thread, and you could go into more detail there too. See - Right now, if someone posts their results i'm not sure if the benchmark relates to the original Logic one, or this new fabfilter/valhalla based test.

 

Plus, one session compared to another which is running on a different plugin architecture, different daw on a different operating system isn't going to be a very reliable comparison. If you got 100 tracks, and someone else had 80 or 120. What does it actually tell you(?) A difference like that could come down to manufacturer drivers, OS System level services, and type of interface the audio device is connected to (PCIe/USB/Thunderbolt etc.).

I'll open up a new thread you are right.

Alright, so I did a test where Plugins would not introduce Sample latency to keep the test purely CPU, here is my result with all the settings:

 

benchmark_new.jpg

 

I've removed L2 because it doesn't fully run on the CPU and it introduces sample latency and no reverb because it's very dynamic and creates CPU overhead.

i got 105 tracks while running 4 more plugs per track tho, with similar settings

are pro tools really that s#!+?

also link the thread after you do it

Alright, so I did a test where Plugins would not introduce Sample latency to keep the test purely CPU, here is my result with all the settings:

 

benchmark_new.jpg

 

I've removed L2 because it doesn't fully run on the CPU and it introduces sample latency and no reverb because it's very dynamic and creates CPU overhead.

 

i got 105 tracks while running 4 more plugs per track tho, with similar settings

are pro tools really that s#!+?

 

also link the thread after you do it

I have a Macbook Pro here now, I'll up you the session I use for comparison which heats up the CPU / SoC the most. It's basically using non-reverb plugins (because they create a CPU overhead based on being very dynamic and then take away load potential) and plugins without sample latency because that offsets load and doesn't go fully on the CPU.

What's with the windows/pro tools posts? ...Am i missing something?

 

Well I'd like to compare it especially with Pro Tools running a powerful desktop processor versus optimized M1 in Logic with the same plugins.

Yes this actually was a good idea to compare 2 daws on 2 different systems with same third party plugins.

This way someone could decide to go with a new Mac and logic or stay on a powerful desktop system in the PC world. Thanks for that but also agree needs its own thread.

https://www.bluecataudio.com/Blog/behind-the-scenes/apple-m1-vs-intel-for-audio-a-benchmark/

from "horse's" mouth, sorry it's third part but i guess relevant anyway