The No.1 Website for Pro Audio
 All  This Thread  Reviews  Gear Database  Gear for sale     Latest  Trending
Logic Pro Multicore Benchmarktest !
Old 8th February 2019
  #3241
Lives for gear
 

Quote:
Originally Posted by Fernand View Post
FWIW, the dual 6 core 3.33 and 3.47 don't really differ noticeably in performance, and the 3.33 is cheaper. As Apple notes, with 6 memory sticks (48 GB) they run a little faster than with 8 sticks (64 gb).
Do you have a source for the memory issue?
I mean, you can even update those 5.1 Mac Pros to 128 GB...
Old 10th February 2019
  #3242
Lives for gear
 

Quote:
Originally Posted by Sascha Franck View Post
Do you have a source for the memory issue?
I mean, you can even update those 5.1 Mac Pros to 128 GB...
It’s not a memory issue at all. It’s just a way for power users to squeeze out a litttle more memory speed. It probably also depends on what apps your working using.

Yes, you can install up to eight 16GB DIMMs in pairs but using only 16GB DIMMS. No 2GB- 4GB- 8GB DIMM pairs mixed with them!
Old 10th February 2019
  #3243
Lives for gear
 

Quote:
Originally Posted by fastlanephil View Post
It’s not a memory issue at all. It’s just a way for power users to squeeze out a litttle more memory speed. It probably also depends on what apps your working using.

Yes, you can install up to eight 16GB DIMMs in pairs but using only 16GB DIMMS. No 2GB- 4GB- 8GB DIMM pairs mixed with them!
Yeah, I'm aware of not mixing DIMM sizes (kinda like a rule for proper performance since almost decades), but I haven't been aware that performance would be better with certain configurations featuring less RAM.
Old 11th February 2019
  #3244
Lives for gear
 
lowkey's Avatar
 

Is there a chart that summarises the results?
Something a bit faster than search through the tread to find results?

Im wondering what sort of step up in performance I might get moving from a 2014 i7 iMac to a new 6 core MBP.
Old 11th February 2019
  #3245
Lives for gear
Quote:
Originally Posted by Sascha Franck View Post
Yeah, I'm aware of not mixing DIMM sizes (kinda like a rule for proper performance since almost decades), but I haven't been aware that performance would be better with certain configurations featuring less RAM.
Here's a link to the Mac Performance Guide for the Westmere Mac Pro's that measured the differences between 2-DIMM, 3-DIMM, and 4-DIMM configurations per socket.
[ macperformanceguide.com: Mac Pro Westmere Real World Memory Bandwidth ]

The upshot is that there's a 2.4-percent and a 2.7-percent loss compared to an optimal 3-DIMM per socket configuration. Also the measurements show that the PC3-10600 vs PC3-8500 throughput is only about a 10-percent performance gain versus the 25-percent clocking difference.

That small clocking performance difference is enough to make me decide to get 6x16GB=96GB of ECC RAM for my 12-core 3.33, upgrading from the 6x8GB=48GB configuration that I have now. (It's all for BlackMagic Fusion 9 Studio, nothing in the audio work requires a RAM upgrade.)
Old 11th February 2019
  #3246
Lives for gear
 

Thanks, MediaGary!
Old 14th February 2019
  #3247
Here for the gear
 
LogicBenchmarks's Avatar
 

Quote:
Originally Posted by lowkey View Post
Is there a chart that summarises the results?
Something a bit faster than search through the tread to find results?

Im wondering what sort of step up in performance I might get moving from a 2014 i7 iMac to a new 6 core MBP.
Yes, there is. You can find it here.
Old 14th February 2019
  #3248
Lives for gear
 
lowkey's Avatar
 

^awesome thanks!!
Old 14th February 2019
  #3249
Lives for gear
Quote:
Originally Posted by Fernand View Post
...
You can get decommissioned top notch HP Server RAM for it on e-bay.
First look for "mac Pro RAM", they usually hit you with a "mac brand
surcharge". Look at the exact type, and search again for them without
mentioning "mac". You can usually find 32GB for ~ $50.
As an example of the influence these kinds of posts can have, looked at my 48GB Mac Pro and decided to go to 128GB would be a dandy upgrade to assist my video editing ambitions and disk caching ambitions.

As you know, the higher density DIMMs carry a price premium, and there is a further premium for Mac-compatible RAM, even though it's a matter of matching up specifications that are conveniently hidden from us by the Mac RAM vendors.

After some reading around in MacRumors, and some detective work, I figured out the the "L" for low-voltage RAM is simply "better" RAM that would work at 1.35v and at 1.5v as in our Mac Pro machines. The 2Rx4 specification is critical to have. I bought some Samsung-manufactured RAM from a "make an offer" vendor on eBay. They have an HP part number. I put in the 8 sticks, and the Mac Pro recognized 6 of them. So the bad news is that it's "only" 96GB.

The good news that someone here can benefit from buying the two unused DIMMs at $50 for the pair. PM me for details. The spec of what I bought is below.

Samsung 16GB 2Rx4 M393B2G70AH0-YH9Q5 PC3L-10600R DDR3 REG ECC HP P/N 628974-081

I'll put up an article soon with the Geekbench results.
Old 15th February 2019
  #3250
Lives for gear
 

Quote:
Originally Posted by LogicBenchmarks View Post
Yes, there is. You can find it here.
Fwiw, just looked at the results - kinda weird as I can run 227 tracks (at buffersize 32 samples - but that really only affects the live thread and the test isn't too tough regarding that) on my Mac Pro 12-Core 2.66 Mid 2010, which is one more than what the Mac Pro 12-Core 3,06 Mid 2012 shows.
I do have my Process Buffer Range set to "large", though.
With the "small" setting things go down to 200 tracks, using "medium" it's 223.
Old 24th February 2019
  #3251
Here for the gear
 

Hi there

HP z820 dual Xeons e52670 with 64 gb Ram (bought a year ago for 1000 euros) running High Sierra 10.13.6

255 tracks

Buffer set to 32 samples
Threads set to 32 (all i got)
Process Buffer Range set to small. (if set to large the overhead is around 40%)

Shared album - Theofilos Mitrentsis - Google Photos
Old 25th February 2019
  #3252
Lives for gear
 

Quote:
Originally Posted by amadeusex View Post
Hi there

HP z820 dual Xeons e52670 with 64 gb Ram (bought a year ago for 1000 euros) running High Sierra 10.13.6

255 tracks

Buffer set to 32 samples
Threads set to 32 (all i got)
Process Buffer Range set to small. (if set to large the overhead is around 40%)

Shared album - Theofilos Mitrentsis - Google Photos
So, Hackintoshs still are the way to go it seems...
Old 17th March 2019
  #3253
Here for the gear
 
LogicBenchmarks's Avatar
 

Quote:
Originally Posted by Sascha Franck View Post
Fwiw, just looked at the results - kinda weird as I can run 227 tracks (at buffersize 32 samples - but that really only affects the live thread and the test isn't too tough regarding that) on my Mac Pro 12-Core 2.66 Mid 2010, which is one more than what the Mac Pro 12-Core 3,06 Mid 2012 shows.
I do have my Process Buffer Range set to "large", though.
With the "small" setting things go down to 200 tracks, using "medium" it's 223.
That's a lot for your Mac. I have the Mid 2010 2,93 12-core and only get around 175 tracks. I don't understand why, because all other people get much more:
Mac Pro 12-Core 2,93 Mid 2010 – Logic Benchmarks
(my score is the one at the bottom, click it to see more details)

I got that with the Buffer at 32. With the Buffer set to 1024 I could barely pass 150 tracks.

Also, look at the pics I attached: One thread stayed very low, that's when I tried it with Buffer set to 1024. When I set it to 32 it got higher, but still not to its fullest. I wonder if something is wrong with my Mac. Maybe that would explain why my score is so low compared to others.
Attached Thumbnails
Logic Pro Multicore Benchmarktest !-2.png   Logic Pro Multicore Benchmarktest !-ohne-titel.png  
Old 17th March 2019
  #3254
Lives for gear
 

Quote:
Originally Posted by LogicBenchmarks View Post
That's a lot for your Mac. I have the Mid 2010 2,93 12-core and only get around 175 tracks. I don't understand why, because all other people get much more:
Mac Pro 12-Core 2,93 Mid 2010 – Logic Benchmarks
(my score is the one at the bottom, click it to see more details)

I got that with the Buffer at 32. With the Buffer set to 1024 I could barely pass 150 tracks.

Also, look at the pics I attached: One thread stayed very low, that's when I tried it with Buffer set to 1024. When I set it to 32 it got higher, but still not to its fullest. I wonder if something is wrong with my Mac. Maybe that would explain why my score is so low compared to others.

As said, I think for my tests I had the process buffer set to large. Makes quite a difference. The used interface might add something to the scenario as well.

Regarding your screenshot of the CPU meter, that's just normal. The rightmost bar is the "live thread", which is exclusively reserved for all record-enabled tracks and hence not sharing any workload (which is what the others do).
Old 17th March 2019
  #3255
Here for the gear
 
LogicBenchmarks's Avatar
 

Quote:
Originally Posted by Sascha Franck View Post
As said, I think for my tests I had the process buffer set to large. Makes quite a difference. The used interface might add something to the scenario as well.

Regarding your screenshot of the CPU meter, that's just normal. The rightmost bar is the "live thread", which is exclusively reserved for all record-enabled tracks and hence not sharing any workload (which is what the others do).
Ahh thanks, I learned something.

Large buffer did the opposite for me. Audio Interface didn't change much however.
Old 5th May 2019
  #3256
Quote:
Originally Posted by Sascha Franck View Post
As said, I think for my tests I had the process buffer set to large. Makes quite a difference. The used interface might add something to the scenario as well.

Regarding your screenshot of the CPU meter, that's just normal. The rightmost bar is the "live thread", which is exclusively reserved for all record-enabled tracks and hence not sharing any workload (which is what the others do).
I've just tested my 2018 Mac Mini i5 today and submitted the results to LogicBenchmarks's website. What I discovered is the same as he referred to in that the last thread is idle at higher buffer settings (128+) but gets more active the lower you go. At 32 buffer it was hitting 60-70% of maximum in Logic's thread processing display. Interestingly enough, I could run a few extra tracks at the lower buffers. If memory serves, I got 131 tracks at 128, 133 tracks at 64 and 135 at 32.

You are correct with regard to the rightmost bar being reserved for realtime performance, which I assume is also the reason why the lowest buffers activates it.
Old 7th May 2019
  #3257
Lives for gear
Quote:
Originally Posted by LogicBenchmarks View Post
I just did the new test on the 7980xe hack (4.5ghz). I can get all 255 tracks to play at 44.1 with no problems. At 88.2 I could get about 230, but when I wen't to do a screenshot I got the CPU error. I dropped it down to 215 and I could grab the screenshot and browse the web so I would consider it stable somewhere around there.

This isn't the Evan test (I can get all 255 tracks at 192 in that one). It's this one-

https://www.logicprohelp.com/forum/v...c.php?t=138612
Attached Thumbnails
Logic Pro Multicore Benchmarktest !-screen-shot-2019-05-06-6.54.03-pm.jpg  
Old 12th May 2019
  #3258
Lives for gear
 
JblKid95's Avatar
 

MacBook pros have throttling so this test won’t be accurate in that regard
Old 29th May 2019
  #3259
Here for the gear
 

Hyperthreading not working

I'm getting much lower results than expected on my 2018 13” i7 MacBook Pro, because it appears that Logic will only use four of my eight (including hyperthreading) cores. It’ll max those four cores out, but the other four show no activity in Activity Monitor’s CPU Usage panel, iStat shows that CPU activity maxes out at 50%, and I get terrible benchmark results.

Weirdly, this only happens with Logic (Live and Studio One use all eight cores), and booting into Safe Mode solves it (ie, all cores are clearly being used), so I’m wondering if it might be a coreaudiod problem.

In this benchmark - https://logicbenchmarks.com - I get about 70 tracks in regular macOS with only four cores showing any action, but 120 in Safe mode with all eight cores doing their thing. Logic’s own CPU meter shows all eight (well, seven plus one for input) cores maxing out, but it’s obviously ‘lying’, because Activity Monitor only shows four cores working, overall CPU usage caps out at 50% with Logic showing less than 400% usage, and the benchmark result is very low. And, like I say, it all works properly in Safe Mode, the only apparent variable within Logic itself being that the no audio drivers apart from FaceTime Audio are available.

Curious as to whether this is a general issue or only a problem for me. Can anyone confirm similar behaviour with Logic Pro X 10.4.4 in macOS X 10.14.5?

Thanks.
Old 30th May 2019
  #3260
Lives for gear
 
BezowinZ's Avatar
Hope to see results for the new iMacs as I'm strongly considering a purchase in another day or so.

With normal use, my 2.8GHz i7 MBP with 16GB RAM is always using 50-65% of CPU in 30-40 Instrument Track templates and only 10-15 of those active. And the fan starts weezing at around 60%.

Folks' test results seem to crush my real life experience. I may get curious enough to test it myself.
Old 30th May 2019
  #3261
Quote:
Originally Posted by S_Bruv View Post
I'm getting much lower results than expected on my 2018 13” i7 MacBook Pro, because it appears that Logic will only use four of my eight (including hyperthreading) cores. It’ll max those four cores out, but the other four show no activity in Activity Monitor’s CPU Usage panel, iStat shows that CPU activity maxes out at 50%, and I get terrible benchmark results.

Weirdly, this only happens with Logic (Live and Studio One use all eight cores), and booting into Safe Mode solves it (ie, all cores are clearly being used), so I’m wondering if it might be a coreaudiod problem.

In this benchmark - https://logicbenchmarks.com - I get about 70 tracks in regular macOS with only four cores showing any action, but 120 in Safe mode with all eight cores doing their thing. Logic’s own CPU meter shows all eight (well, seven plus one for input) cores maxing out, but it’s obviously ‘lying’, because Activity Monitor only shows four cores working, overall CPU usage caps out at 50% with Logic showing less than 400% usage, and the benchmark result is very low. And, like I say, it all works properly in Safe Mode, the only apparent variable within Logic itself being that the no audio drivers apart from FaceTime Audio are available.

Curious as to whether this is a general issue or only a problem for me. Can anyone confirm similar behaviour with Logic Pro X 10.4.4 in macOS X 10.14.5?

Thanks.
Is Logic set to use all processing threads in the Audio preferences?
Old 30th May 2019
  #3262
Quote:
Originally Posted by BezowinZ View Post
Hope to see results for the new iMacs as I'm strongly considering a purchase in another day or so.

With normal use, my 2.8GHz i7 MBP with 16GB RAM is always using 50-65% of CPU in 30-40 Instrument Track templates and only 10-15 of those active. And the fan starts weezing at around 60%.

Folks' test results seem to crush my real life experience. I may get curious enough to test it myself.
The test is not ideal as a "typical user experience" indicator. I've mentioned this to LogicBenchmarks and he might make an alternative test which makes use of buss processing, audio tracks etc. for a more complex test that would be a better approximation of real world use.

As I might have mentioned already I got the highest track count with a buffer of 32 samples, which does not work well when I'm producing. (Although you can get around the problem to a certain degree by selecting a track which does not trigger Logic's "Live"/realtime mode, pressing play and only then select the track you want to edit. Annoying, though!)
Old 30th May 2019
  #3263
Here for the gear
 

Quote:
Originally Posted by thermal View Post
Is Logic set to use all processing threads in the Audio preferences?
Yes it is. I've actually now tried this on several current Macs (two MBPs and an iMac Pro), and they all do the same thing, so it's obviously a rather major bug, which I've reported to Apple.
Old 5th June 2019
  #3264
Gear Addict
 

Hey Guys,
I have a question for you. i have a 2013 6 core macpro @3,5Ghz. I do mainly music production and mix using vepro and a lot of vst instruments. I record using 1 channel or 2 at a time. I'd like to get my Logic Pro a bit faster with more power for VSTs and mix plugins. And also faster exports/bounces. Is it worth upgrading CPU to a 12 core @2,7Ghz ? But I know logic has single core CPU spike and I'm afraid downgrading each of them to 2,7GHz...what do you think ? thx
Old 6th June 2019
  #3265
Here for the gear
 
LogicBenchmarks's Avatar
 

The new Mac Pro can play 1000 tracks!

https://logicbenchmarks.com/2019-mac...ftware-tracks/
Old 7th June 2019
  #3266
Gear Addict
 

Quote:
Originally Posted by LogicBenchmarks View Post
The new Mac Pro can play 1000 tracks!

https://logicbenchmarks.com/2019-mac...ftware-tracks/
What we need is Logic balance well CPU activity between cores. We don't want CPU spikes on 1 core when adding plugins or bus. Do you have an idea about my question below ? thanks a lot !!

Quote:
I have a question for you. i have a 2013 6 core macpro @3,5Ghz. I do mainly music production and mix using vepro and a lot of vst instruments. I record using 1 channel or 2 at a time. I'd like to get my Logic Pro a bit faster with more power for VSTs and mix plugins. And also faster exports/bounces. Is it worth upgrading CPU to a 12 core @2,7Ghz ? But I know logic has single core CPU spike and I'm afraid downgrading each of them to 2,7GHz...what do you think ? thx
Old 14th June 2019
  #3267
Gear Nut
 
Trendkill's Avatar
 

https://support.apple.com/en-us/HT203718

Start your engines!
1000 Instruments in Logic Pro X 10.4.5
Old 15th June 2019
  #3268
Gear Maniac
 
sharky's Avatar
Quote:
Originally Posted by Lefrançais View Post
Hey Guys,
I have a question for you. i have a 2013 6 core macpro @3,5Ghz. I do mainly music production and mix using vepro and a lot of vst instruments. I record using 1 channel or 2 at a time. I'd like to get my Logic Pro a bit faster with more power for VSTs and mix plugins. And also faster exports/bounces. Is it worth upgrading CPU to a 12 core @2,7Ghz ? But I know logic has single core CPU spike and I'm afraid downgrading each of them to 2,7GHz...what do you think ? thx
I'm curious about this as well. I understand there are ways to change your busses to sort of "force" Logic to use other cores and load balance, but I'm curious about real world practical performance of less cores and higher clock speed vs. more cores and lower clock speed.
Old 1 week ago
  #3269
Quote:
Originally Posted by sharky View Post
I'm curious about this as well. I understand there are ways to change your busses to sort of "force" Logic to use other cores and load balance, but I'm curious about real world practical performance of less cores and higher clock speed vs. more cores and lower clock speed.
I have the 12 core @2.7Ghz. I can run over 200 Virtual instruments at once and even 1000 audio tracks. (The only reason I couldn't run more audio instruments is because I ran out of Ram. 64GB.
Old 6 days ago
  #3270
Lives for gear
 
Chevron's Avatar
 

Quote:
Originally Posted by chevaliermusic View Post
I have the 12 core @2.7Ghz. I can run over 200 Virtual instruments at once and even 1000 audio tracks. (The only reason I couldn't run more audio instruments is because I ran out of Ram. 64GB.
Is that for this benchmark test with your 12 core? The results indicate you should get more... https://logicbenchmarks.com/benchmark-results/
Post Reply

Welcome to the Gearslutz 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…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump