Quantcast
Logic Pro Multicore Benchmarktest ! - Page 111 - Gearspace.com
The No.1 Website for Pro Audio
Logic Pro Multicore Benchmarktest !
Old 19th November 2020
  #3301
Here for the gear
 
I just did this benchmark on my 2017 iMac (4.2 i7) and the new M1 MacBook Air. Once I figured out the weirdness with processing threads, I got a score of 145 on the iMac, 153 on the MacBook Air natively, and 130 on the MBA under Rosetta. Both using 64 sample buffers through built-in audio (though weirdly, changing the buffer size didn't seem to impact the score - also, the same buffer size on the MBA produces a significantly lower RTL than the iMac.)

I need to try some other tests... though that result is amazing considering it's a passively cooled laptop, it's not in line with the difference in these two machines' Geekbench scores, which shows the MBA blowing my iMac out of the water in every way by significant margins. The small difference between Rosetta and Native modes is also a head scratcher.

EDIT: I just performed the test with the "NewLogicBenchmarkTest" file that is elsewhere on Gearslutz (the one with Sculpture instead of the FM synth), and the scores, under similar buffer settings, were 70 for my iMac, and 99 tracks for the MacBook Air running native ARM. The iMac's fan really got going on this test, and the score on the iMac would lower when the fan was working harder, but I couldn't get the MacBook Air to throttle back. And after several repetitions, it was only warm to the touch.

EDIT 2: On that same newer test, under Rosetta, LPX can do 60 tracks reliably on the MacBook Air. Though it was a little more murky with the MacBook Air than the iMac finding the cutoff point where it could reliably keep going. The MBA would play a few bars on up to 70 tracks before overloading.
Old 19th November 2020 | Show parent
  #3302
plx
Lives for gear
 
🎧 10 years
Quote:
Originally Posted by Nathanael D. ➡️
I just did this benchmark on my 2017 iMac (4.2 i7) and the new M1 MacBook Air. Once I figured out the weirdness with processing threads, I got a score of 145 on the iMac, 153 on the MacBook Air natively, and 130 on the MBA under Rosetta. Both using 64 sample buffers through built-in audio (though weirdly, changing the buffer size didn't seem to impact the score - also, the same buffer size on the MBA produces a significantly lower RTL than the iMac.)

I need to try some other tests... though that result is amazing considering it's a passively cooled laptop, it's not in line with the difference in these two machines' Geekbench scores, which shows the MBA blowing my iMac out of the water in every way by significant margins. The small difference between Rosetta and Native modes is also a head scratcher.

EDIT: I just performed the test with the "NewLogicBenchmarkTest" file that is elsewhere on Gearslutz (the one with Sculpture instead of the FM synth), and the scores, under similar buffer settings, were 70 for my iMac, and 99 tracks for the MacBook Air running native ARM. The iMac's fan really got going on this test, and the score on the iMac would lower when the fan was working harder, but I couldn't get the MacBook Air to throttle back. And after several repetitions, it was only warm to the touch.

EDIT 2: On that same newer test, under Rosetta, LPX can do 60 tracks reliably on the MacBook Air. Though it was a little more murky with the MacBook Air than the iMac finding the cutoff point where it could reliably keep going. The MBA would play a few bars on up to 70 tracks before overloading.
LOL.
my Mini i7 gets 83 Stable (i need to open it up and dust it tho).

99 is absolutely insane for a passively cooled design. (it's better than an 8-core i9 16"!)
Old 19th November 2020 | Show parent
  #3303
Gear Head
 
Niclas.G's Avatar
 
Wow. This is insane. Just did that test with my imac 27” 4.0 2014 and it scored 54! Its mostly good enough for my studio yet, and to know i can go for a AIR next time and it will perform TWICE AS GOOD is just. Breath taking!
Old 24th November 2020 | Show parent
  #3304
Lives for gear
 
BobTheDog's Avatar
 
🎧 10 years
Mac Mini M1 16GB with original project from first post, 44.1K 128 buffer:

Old 25th November 2020 | Show parent
  #3305
plx
Lives for gear
 
🎧 10 years
Quote:
Originally Posted by BobTheDog ➡️
Mac Mini M1 16GB with original project from first post, 44.1K 128 buffer:

do the NewLogicBenchmark instead please
Old 25th November 2020 | Show parent
  #3306
Lives for gear
 
BobTheDog's Avatar
 
🎧 10 years
Quote:
Originally Posted by plx ➡️
do the NewLogicBenchmark instead please
This one?

Anyone interested in my new benchmark to replace the dated Evan's test?
Old 25th November 2020 | Show parent
  #3307
Lives for gear
 
🎧 15 years
Yeah TNM's is the new one.
Old 25th November 2020 | Show parent
  #3308
Lives for gear
 
🎧 15 years
Also with the TNM test- the last track (the audio track) is there so you can select it and NOT auto arm any of the tracks. I'm not sure if it makes a difference in the newest tests or 10.5, but these benchmarks are only supposed to measure playback tracks.

IOW nothing should be armed, LPX has to shift the armed track to the low latency 'recording' engine.


It would be nice to figure out a benchmark test that can test low latency performance. We all know about processor spikes during tracking- at some point, that becomes more relevant that pure playback performance for many of us.
Old 25th November 2020 | Show parent
  #3309
Lives for gear
 
🎧 15 years
Quote:
Originally Posted by RyanC ➡️
It would be nice to figure out a benchmark test that can test low latency performance.
I posted a project doing a decent job for that in the other thread. And at least 3 people have already posted their results, one of them owning a new MB Air. Results were even more stunning IMO. These new machines kill it with their single core performance.
Old 25th November 2020 | Show parent
  #3310
Lives for gear
 
🎧 15 years
Quote:
Originally Posted by Sascha Franck ➡️
I posted a project doing a decent job for that in the other thread. And at least 3 people have already posted their results, one of them owning a new MB Air. Results were even more stunning IMO. These new machines kill it with their single core performance.
Cool, which thread? Or can you link the post? Curious to try it
Old 25th November 2020 | Show parent
  #3311
Lives for gear
 
🎧 15 years
Quote:
Originally Posted by RyanC ➡️
Cool, which thread? Or can you link the post? Curious to try it
Well, my test project can be found in the new benchmark thread and in the M1 performance one where you'll also find some results. This is the page where you'll find some reports (basically a 16-core nMP vs. a MB Air - the latter taking the trophy with ease...).
M1 Apple Silicon Experiences in Audio
And here's the direct link to the test project (description included):
https://drive.google.com/file/d/10IG...ew?usp=sharing
Old 25th November 2020
  #3312
Lives for gear
 
🎧 15 years
Seriously, those M1 machines simply kill in terms of single core performance. And in case you're not into huge orchestral arrangements, it should be pretty much all you need for most other things (minus the connectivity issues - regarding that, the situation is as poor as it ever was with Apples low end models).
Old 26th November 2020 | Show parent
  #3313
Lives for gear
 
🎧 15 years
Quote:
Originally Posted by Sascha Franck ➡️
Seriously, those M1 machines simply kill in terms of single core performance. And in case you're not into huge orchestral arrangements, it should be pretty much all you need for most other things (minus the connectivity issues - regarding that, the situation is as poor as it ever was with Apples low end models).
Yeah, it's impressive. And this is the Air getting 45 space designers at 32- do we have a mini or MBP test yet? Should be around ~50?

I only get ~25 at 32 sample buffer on my 7980xe hacks, but I get around 120 space designers at 128 buffer, and ~310 tracks in the TNM test. These computers really struggle at 32 in logic especially. My biggest issue with these hacks is I would give up some playback performance for better low latency performance.

It seems like a the M1 but be a bit too thin there, because I could only really sacrifice maybe 1/3 of the playback performance before I'd have to open current projects and start freezing stuff- and it looks like the M1's are slightly less than half. But the rumored 12c (8+4) M1x would probably do it for me...funny to think about 4 mac book pro's or imac's in the machine room- but I might just do it if we are going to be waiting a while for a asMP.

I'll be looking forward to performance tests with 3rd party plugins running without rosetta...
Old 26th November 2020 | Show parent
  #3314
Lives for gear
 
🎧 15 years
Quote:
Originally Posted by RyanC ➡️
My biggest issue with these hacks is I would give up some playback performance for better low latency performance.
Really? I mean, is that your personal experience?
Usually, with Logic it shouldn't be like that. I mean, it's using a higher buffer size (1024 samples) for playback tracks regardless of your settings in the audio preferences dialog, so the only tracks affected by lower set buffersizes are those record enabled.
At least that's my experience (and that of plenty of other folks). I can run pretty much every bit as many tracks in the "normal" Logic benchmark tests at either 32, 64 or 128 samples buffersize.

But then, there *might* be some side effects of lower buffer settings in a typical daily working scenario, such as in using sends/busses for both the playback and "live" tracks.
Maybe it'd be interesting to have another test project, taking both multithread and low latency singlethread performance into account. I might think about something...
Old 26th November 2020 | Show parent
  #3315
Lives for gear
 
🎧 15 years
Quote:
Originally Posted by Sascha Franck ➡️
Really? I mean, is that your personal experience?
Usually, with Logic it shouldn't be like that. I mean, it's using a higher buffer size (1024 samples) for playback tracks regardless of your settings in the audio preferences dialog, so the only tracks affected by lower set buffersizes are those record enabled.
At least that's my experience (and that of plenty of other folks). I can run pretty much every bit as many tracks in the "normal" Logic benchmark tests at either 32, 64 or 128 samples buffersize.

But then, there *might* be some side effects of lower buffer settings in a typical daily working scenario, such as in using sends/busses for both the playback and "live" tracks.
Maybe it'd be interesting to have another test project, taking both multithread and low latency singlethread performance into account. I might think about something...
I think I didn't explain well. What I mean is my current machines have more playback ability than I need, I would sacrifice some of that for a machine that has more low latency capability- but only to a point. I don't like having to freeze tracks and some future plugins will likely use more cpu.

So currently I'm at TNM: 310 tracks, Your test (32):25. For what I do, TNM:~200 Your test (32):50 would likely be a worthwhile upgrade, even if it looses some ability in the playback metric. But I don't think TNM: ~100 your test:50 would be enough muscle for me on the playback side.

M2/M1x looks like it might be there for me with the 12c, 8+4 setup. Hoping there's a m2/m1x mini with that and 10gbe.
Old 26th November 2020 | Show parent
  #3316
Lives for gear
 
🎧 15 years
Quote:
Originally Posted by RyanC ➡️
I think I didn't explain well. What I mean is my current machines have more playback ability than I need, I would sacrifice some of that for a machine that has more low latency capability- but only to a point. I don't like having to freeze tracks and some future plugins will likely use more cpu.
I'm pretty much in a similar camp - just that I don't even exactly need the "bandwidth" of my current Mac Pro most of the time (I'm not doing larger scaled things and also don't think I will in the foreseeable future). But I would pretty much like plenty of "realtime juice" - which seems to be exactly the main strength of the current M1 lineup.
Hence, regarding both multithread and singlethread performance, any of these new machines would suit me just fine, my only gripe being the rather miserable connectivity situation. I'd really like to get a Macbook (once gigging is at least somewhat back so I can afford it), but I'd have to spend quite some money regarding connections to what I need in a stationary situation (in real mobile situations my demands are pretty low).
Old 31st December 2020 | Show parent
  #3317
Here for the gear
 
🎧 5 years
Question

Has anyone with an M1 Air encountered any noticeable thermal throttling when pushing the processor beyond 70% utilization for extended periods of time? Since audio production doesn't tend to tax the GPU cores, I'm wondering if the Air vs Pro is almost a moot point for audio work. I've seen plenty of gaming and prolonged Cinebench tests reveal the superiority of the Pro's cooling solution, but it usually doesn't result in much more than about 15% throttled performance. I suppose if I was primarily an Ableton Live user, the Pro would be a safer bet since its real-time audio engine tends to be a bit more CPU sensitive. I wonder if running this benchmark on a loop for 10+ minutes might reveal the Air's limitations.
Old 31st December 2020 | Show parent
  #3318
Here for the gear
 
🎧 5 years
Quote:
Originally Posted by BobTheDog ➡️
Mac Mini M1 16GB with original project from first post, 44.1K 128 buffer:

Impressive result. Despite all using the same chip, the Mini really does seem to have that extra bit of grunt over the laptops. I was able to borrow a friend's base model MacBook Pro and here were the results using Evan's original benchmark:

M1 13" MacBook Pro (8GB RAM, 256GB SSD) / Logic Pro 10.6.1
I/O Buffer Size: 128 samples / Processing Buffer Range: Medium
  • CPU: Automatic = 92 tracks
  • CPU: 8 Processing Threads = 148 tracks
  • Processing Buffer Range: Large = 162 tracks
Logic seems to behave in some counterintuitive ways on this M1. Unlike on Intel systems performance appears to be worse at I/O Buffer Sizes of 256 samples and higher, compared to 128 samples and below. Also, on the M1 MacBook Pro with Logic’s performance meter maxing out all processing threads for a continuous, uninterrupted 10 minute loop I still couldn’t get the fan to spin up. Considering music production doesn't typically max all CPU/GPU cores like video encoding or 3D rendering, the Air might be just as viable as the Pro.
Old 1st January 2021 | Show parent
  #3319
Lives for gear
 
BobTheDog's Avatar
 
🎧 10 years
Quote:
Originally Posted by Tronam ➡️
Impressive result. Despite all using the same chip, the Mini really does seem to have that extra bit of grunt over the laptops. I was able to borrow a friend's base model MacBook Pro and here were the results using Evan's original benchmark:

M1 13" MacBook Pro (8GB RAM, 256GB SSD) / Logic Pro 10.6.1
I/O Buffer Size: 128 samples / Processing Buffer Range: Medium
  • CPU: Automatic = 92 tracks
  • CPU: 8 Processing Threads = 148 tracks
  • Processing Buffer Range: Large = 162 tracks
Logic seems to behave in some counterintuitive ways on this M1. Unlike on Intel systems performance appears to be worse at I/O Buffer Sizes of 256 samples and higher, compared to 128 samples and below. Also, on the M1 MacBook Pro with Logic’s performance meter maxing out all processing threads for a continuous, uninterrupted 10 minute loop I still couldn’t get the fan to spin up. Considering music production doesn't typically max all CPU/GPU cores like video encoding or 3D rendering, the Air might be just as viable as the Pro.
I only managed to get to 196 by adding one at a time, if you stop and play at 196 it would overload. For totally stable stop and play I think 188 was the maximum.

For me buffer size made no difference at all on any setting!
Old 30th January 2021
  #3320
Gear Nut
 
steverino's Avatar
 
🎧 10 years
have upgraded from a 2012 Mini i7 16GB RAM and can say the M1 Mini is a winner... it screams...
Old 30th January 2021
  #3321
plx
Lives for gear
 
🎧 10 years
I get 108 tracks on Logic TNM with M1 13" Pro
My Mac Mini i7 (Which i already sold) got 78 (80 on a good day).
Old 30th January 2021 | Show parent
  #3322
Here for the gear
 
🎧 5 years
Quote:
Originally Posted by plx ➡️
I get 108 tracks on Logic TNM with M1 13" Pro
My Mac Mini i7 (Which i already sold) got 78 (80 on a good day).
And for reference I get around 125 tracks on my 2019 Intel i9 8c/16t iMac. Not a bad showing for the M1, especially considering it's only their first bottom tier entry level CPU.
📝 Reply

Similar Threads

Thread / Thread Starter Replies / Views Last Post
replies: 295 views: 76967
Avatar for anguswoodhead
anguswoodhead 26th March 2013
replies: 1188 views: 195855
Avatar for jonljacobi
jonljacobi 15th June 2019
replies: 1296 views: 188445
Avatar for heraldo_jones
heraldo_jones 1st February 2016
replies: 29967 views: 3247567
Avatar for jBranam
jBranam 14 hours ago
Post Reply

Welcome to the Gearspace Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…

Forum Jump
Forum Jump