2019 Mac Pro with Logic

[Dewdman42]

I have to say, even though I wish Apple had made an i9 based macpro that i think would be better for music production, I personally would still pick a macpro over an iMac if I were going to buy one today, precisely because of the fact I want everything inside the box on pci slots.

 

Plus you can use superior gpu and whatever collection of monitors you want

 

The iMac has only slightly better single core performance then the Mac Pro but the 12 core macpro kicks the i9’s butt on multicore. Single core is ultimately more important for low latency audio while tracking. So yes the i9 will do better for that. But the macpro is still pretty capable. My main beef with the nMP is the price.

 

On that note I would have thought your mbp Would have great single core performance compared to the 5,1. But probably not the greatest multicore. Tell us the specs so we can make a proper comparison.

[Ploki]

I mean... Someone played back 100 tracks with 32 buffer size on the mac pro?

and if you want really low latency just get an interface with basic DSP and track with direct monitoring.

 

There's literally nothing in iMac that would have me pick it over a Mac Pro

 

Ultimately for audio, stable performance will be more important than burst speeds... So having a CPU that consistently delivers is better than one that craps out because of heat in 15 minutes. and this is where the Mac Pro excels at.

[Dewdman42]

Heat is an issue on the iMac, but not the i9, for the record.

 

The new macpro runs very quiet and cool but people are complaining that the turbo mode is not ramping up very well in various daw tests. There are some interesting threads on gearslutz about it.

 

I still do not think the nMP is optimal for us in audio due to xeons. I just think it’s currently the best Apple branded choice, but also very overpriced. It’s why I plan to stay on my 5,1 for a few more years. But hey 100 tracks with a 32 buffer AND IF THEY ARE TRYING TO RECORD AT LEAST ONE TRACK AT THE SAME TIME, is the test to make.

 

It’s hard to get good information on this topic because every owner appears to be the biggest biased fan of whatever computer they own and they will produce all manner of positivity about it. But a truly objective comparison of different platforms is hard to come by.

[Ploki]

tbh the only thing holding me back from a Mac Pro is price, else im totally up for it instead of the mini.

[angelonyc]

This thread has become very informative. When I bought my 2015 mac pro, no one at Apple could tell me how many cores to get.. I gave specs of wanting 40 live virtual instruments. I got answers from the most minimal system to only the 12 core top of the line. Problem is some VI's use exponentially more power than others.

 

I think I need a 2019 mac pro, cause I like to keep VI's live while composing. cause I'm constantine jumping to different instruments, making minor adjustments in notes, timing, velocity, duration etc. But I'm in sticker shock over the price. And you know 3rd party companies will jump to eat up CPU power to pack more features into their plug-ins. It's a never ending sea-saw. And 6 years ago, I would have never have dreamed of having over 100 tracks in a piece.

 

Yah I heard that argument, if you need that many tracks, you don't know how to arrange..

[Makzimia]

@Dewdman42 Thank you for the comparisons, most useful!. I’m on my 2018 MBP now after stupidly giving up on my 2012 5.1 Mac Pro. It’s amazing how little the MBP handles compared to my 5,1, even though I’m now on thunderbolt 3 instead of FireWire for example. Definitely suspect single core speed as the issue. Logic Pro decides regularly to spit the dummy out on a very light project. I was going to go the new Mac Pro, lovely wife was fine.., I’m not. Leaning more and more to i9 iMac (pro?). I just really wanted drive bays again too...

 

What are the specs of your 2018 MBP?

 

In my sig mate . only thing missing there is the 1tb SSD.

[Dewdman42]

I can't find a MBP on geek bench that matches that spec. Which of the following most closely matches your machine:

 

https://browser.geekbench.com/macs/433

 

I see 6 cores at 2.2 or 2.6ghz. I don't see a 2.4ghz model, at least on Geekbench...

[Dewdman42]

In any case, both of those models have pretty weak multi-core scores. So its no wonder you are mixing less then your old 5,1. Though your MBP should be able to handle quite a bit more then the 5,1 in terms of what you can record to a track, or what plugins you can have in the signal path for a track while recording it in live mode...with a lower latency setting...then what you could do with the 5,1. But overall, I can see what maybe you're running out of mixing juice...your multicore score is something around 4800-4900. The 5,1, with top end CPU is 30% higher. But you single core performance is 30% higher then the 5,1 was.

[Dewdman42]

Use this page to see where your computer ranks compared to many other mac models:

 

https://browser.geekbench.com/mac-benchmarks

 

I guess you're single core performance is only 10-20% lower then the very top fastest mac available.

 

Your multicore though, is 1/3 the score of a 12 core nMP and top end iMac. 30% lower then your old 5,1 too.

[Makzimia]

@Dewdman42 it’s a 2.6ghz, the ram is slower though, 2400 I think. You were pointing out that bottle neck earlier. I’ve removed that part of sig, I can see how that was confusing, sorry.

[Makzimia]

On the subject of single core, that’s where my MBP is choking on a 128 buffer which is what I ran my 5,1 on, is why it chokes.. I get the change setting notice as it stops. It’s very random of when. I’ve seen the single core meter spike then bam. It’s a very light song with minimal automation and plugins that ran fine on the 5,1. I’m just really sad I jumped and lost on the way. Guess the dual xenon 3.46 ghz CPUs in the 5,1 really were that good, even though I was on 7200rpm spinners :-/.

[Dewdman42]

That doesn’t sound normal to me and I think you ought to be able to get better performance out of that mbp. But don’t tell your wife I said that

[Makzimia]

If anyone can point me to what I may need to check on my MBP, would be most appreciative .

[Dewdman42]

could be so many things causing the cpu to spike like that. First, let's check one thing, when you are playing back and not actually tracking anything, do you have an empty dummy track selected to avoid live mode?

 

What are your audio preferences set to?

 

What audio and midi drivers?

 

what else is running on your mac besides your DAW?

 

what plugins are you using when this happens. Are you sure its totally random?

[Makzimia]

Thanks Dewd . I’ll need to check some of that when I’m in studio. But I use a UAD Apollo Quad silverface version with TB3 128 buffers. This is on play back no live tracking. When I do live tracking I’ve had no real issues. It’s random as to when in the song it stops, BUT, it may be near a specific section. I’ll let you know.

 

Thanks

 

Tony

[Dewdman42]

Make sure to create a dummy track... create a midi track, then right click on it and reassign to NO OUTPUT. then make sure to select that track header before hitting play. You should get in the habit of doing that always when not actually recording. That will make sure there are no tracks in live mode. Live mode will attempt to push everything in the signal path of that track through a single core, including AUX's and the master bus too.

 

I would think the UAD stuff would have good drivers and not be a problem there, so unless you have something else funny installed on your system that might spike a core, but I'm more suspicious that you are having a track header selected that has something big in terms of plugins hit a single core at some points.

[Dewdman42]

See this result: viewtopic.php?f=1&t=138612&start=120#p757529

 

The new 16 core Mac pro is crippled with 6 instances of U-HE Diva playing 8 voices each.. too much for single core load so it can only load 6 and leave all the other cores free.. it's the single core performance that is at fault here.. My iMac pro gets 6 also, yet my MacBook pro gets 15.. yes an 8 core MacBook pro gets 15 x 8 voice divas all playing back without logic overloads, vs 6 in a mac pro configured to about 14 grand.. There's a large topic at gearslutz explaining it all (the last 5 or so pages are the ones to read about this from memory), but basically OS X is not allowing the Xeons to keep a steady all core turbo speed and drops below Intel's actual rated speeds, whereas Windows is letting them perform at their maximum. For example in my iMac pro, i get 12 divas in Pro tools/bootcamp vs 6 in Pro tools/Mojave. It's a power limit issue in OSX and we all hope it gets worked out. The reason the MacBook pro gets one per available thread in Logic is because there is enough headroom left over after each one, to put the next one on.. With the mac pro's, the reason they can only play 5 or 6 is because even though those first few instances can be loaded on one core each, every new instance adds overhead even to previous cores, just slightly, but there is no headroom left in the single core performance so it flakes out much quicker. The 9900K gets like 20 or something.

 

Yet, with omnisphere, there is a demo of the new mac pro playing 80 of them and all 16 cores/32 logical cores getting hammered to the max. So really, it's only U-HE synths that seem to not be able to be fully taken advantage of on the Xeons.. I have other heavy cpu synths and I can use them all.. just not Diva.

All the Arturia stuff is fine, Dune 3, the legend which sounds *amazeballs* used about 1/5th the cpu of DIva's moog filter mode. Try the legend, you will love it.. Only 8 voices but he used an AVX trick to get it to perform like other synths with one voice!

[Makzimia]

@Dewdman42 After trying the dummy track and still getting the stop, at the same spot. I then tried my next suspect which was Bassrider from Waves, (never been an issue, but). Anyway, end result.. I then checked my audio settings. I am 99.9% certain I have always had Multithreading set to both Playback and Live, however, I changed to just playback, and I was actually on 256 buffer too, and the issue has gone away. Fact was for whatever reason, it was overloading thread 1 at certain points of the song to over 100% and causing that drop out. Thread 7 I think, in photo, is bearing that burden on Playback, and doesn't even get over 70% now. Problem averted for now, wife breathes a sigh of relief LOL.

 

Thanks for making me look again

 

song shot load

[Dewdman42]

awesome!

[Ploki]

https://www.kvraudio.com/forum/viewtopic.php?f=31&t=538533

 

there's a new diva release with better multicore performance

[9u10m]

Make sure to create a dummy track... create a midi track, then right click on it and reassign to NO OUTPUT. then make sure to select that track header before hitting play. You should get in the habit of doing that always when not actually recording. That will make sure there are no tracks in live mode.

Selecting a MIDI dummy or an audio track will always deactivate one of my cores during playback, here. I can only have all cores working when selecting an instrument track.

[Dewdman42]

I think you mis understand what happens when you select a dummy track. If you have any other track selected that particular track is placed in live mode and forced into a single dedicated core. That core will be more likely to spike or dropout. When you select a dummy track, LogicPro will intelligently use all the cores as it sees fit in the most efficient way possible, also with a large buffer for everything.

[9u10m]

Thanks for your reply. I do get what making sure no tracks are in live mode implies for how LPX handles the core load. Then again, with my 6-core 4930k:

 

- selecting a dummy or audio track > 1 core will display a higher load than the others, 1 core will not work

- selecting an instrument track > 1 core will display a higher load than the others, all cores will work

 

I can replicate 100% in any Logic project, and scenario #1 will always be more likely to give me a CPU overload. In my case it doesn't look like the dummy track trick allows LPX to intelligently use all cores. It's basically going to more or less display the same core load, except one of them will always stop working as soon as I restart playback.

[Ploki]

how about if you manually deselect it by clicking "R", or selecting an audio track

[Dewdman42]

Something is amiss somewhere. By the way the meter in logicpro is showing threads, not cores. If you watch your systemwide cpu meter you can see how the actual cores are being used by those threads ( and other threads from other processes at the same time)

 

In lpx prefs what do you have as the multi threading setting? Auto?

[9u10m]

In my case it does show cores because I deactivated multithreading in the BIOS (my rig is a hackintosh)

 

Processing Threads: either 6 (max) or Automatic, doesn't make a difference after a few playbacks.

Multithreading: Playback and live tracks. Playback tracks only doesn't seem to make much of a difference either.

[Dewdman42]

No. You cannot deactivate multithreading. You may have deactivated “hyper threading”

 

The logicpro meter always shows threads it doesn’t show anything about cores. Threads are moved around from core to core by your operating system. And the rest of your operating system is also making numerous threads and sharing time on those 6 cores also. The meter in logicpro just shows how many threads Logic Pro itself has spawned and by the way some of your plugins may also spawn Multiple threads which logicpro will not know anything about nor show in its thread meter

[9u10m]

Typed too fast, hyper-threading indeed. If it is enabled Logic's performance meter shows 12 threads. Once disabled it only shows 6.

I've had to disable it (since Sierra I think), or my rig won't boot up.

[Dewdman42]

Ok so keep in mind that it is not always beneficial to have more threads running. Logicpro allocates threads as it sees it best. If Logic only allocates 5 threads for whatever the reason, that doesn’t mean you’re only using 5 cores. Watch the system wide cpu meter to see the core use.

[9u10m]

What you're saying makes total sense, although the occasional CPU overloads will be handled better when I select an instrument track (and all 6 threads/cores show as being used in Logic's meter). If using more threads was going to prevent said overloads, then why wouldn't Logic do so?