Max out iMac pro upgrade worth it? Single core score ?’s

[Artcutech]

So I’ve been thinking about taking my 8 core with 64 gb of ram iMac pro and maxing it out to 18 core 128 gb of ram, just watched a video where Linus does this upgrading the cores from 8 to 18 and the ram from the 32 to 128, and everything worked!

My question is, would it even be worth it using Ableton? Even though Ableton supports multi core, it seems some single core aspects are factored in to the cpu usage in Ableton, I can’t tell you how frustrating it is to see activity monitor show only 10 percent cpu being used but Ableton getting ready to break down showing 80 percent usage at a 2048 buffer.

It doesn’t look like the single core scores vary much with the iMac pro, and I know having 18 cores would add some advantage to the cpu in Ableton but would it be more of the same like 5 percent being used in activity monitor while Ableton on the verge of breaking down? I’m assuming this is the case so do you guys think there’s much benefit upgrading?

Also, although I do not currently use logic but have in the past but never tested its limits. Does anyone on here know if logic handles much better under heavy loads? or should I say more in line with what activity monitor is showing? Or is there any Daw that responds to cpu more in what activity monitor shows?

Edit: btw way I’m running live 9.7, is there much improvement with the cpu reflecting more of what’s going on in activity monitor in live 10? Or just any improvements with live 10 and cpu in general?

[fishmonkey]

it really depends on how you work.

if you use lots of tracks then more cores will help. if you tend to have fewer tracks, and especially if some of those tracks are loaded up with CPU-intensive stuff, then single-core performance matters more.

it doesn't make sense to compare the CPU usage in Activity Monitor with the Live audio processing meter, as overall CPU utilisation is not a good measure of DAW performance. you need to think more about what is the weakest link in the chain. if even one of your cores is overloaded, then it doesn't matter if you have 1000 cores available, that one core will glitch out on you...

[Machinesworking]

What fishmonkey wrote. Getting the most powerful computer you reasonably can always helps, but learning the limitations of computers with your work style is more important. It's the difference between knowing that a 2.8ghz 8 core machine is right or if a 3.8ghz six core machine is going to go further for you.

[Artcutech]

thank you fishmonkey and machinesworking for the replies

fish you wrote

"it doesn't make sense to compare the CPU usage in Activity Monitor with the Live audio processing meter, as overall CPU utilisation is not a good measure of DAW performance. you need to think more about what is the weakest link in the chain. if even one of your cores is overloaded, then it doesn't matter if you have 1000 cores available, that one core will glitch out on you..."

this makes perfect sense.

for my work flow I don't think it would make much of a difference as I tend to have fewer tracks that are loaded up with cpu-intensive plug-ins. Apart from my vst to audio bounce, I really hate to freeze and flatten, I use to have to all the time on my MBP and I try to get by without doing so nowadays, but sometimes will if need be.

Also I'm at 24-bit 48k, when switching down to 24-bit 44k it looks to ease up a lil but hardly anything, my interface got sent in for repairs and I'm listening on iMac pro speakers but after just trying it, it sounded slightly more pleasing in a way, maybe my brain is playing tricks on me haha but do you think 44k might add more of a neutral sound as opposed to 48k or is there prob no audible difference?

[TLW]

Personally I’d rather have fewer faster cores, assuming I have enough to manage in the first place, for the reasons fishmonkey gives. I’m contemplating getting a new iMac or iMac Pro after Catalina settles down and at the moment I think I’d go with the standard base-model cpu spec. on the Pro. Though I’m not sure it would beat an i9 iMac or even justify the extra cost over a current i7 for what I do.

As for the 48KHz vs 44.1KHz thing I doubt there’s any audible difference unless having a frequency ceiling of about 24KHz means some aliasing is pushed up into audio frequencies where you can’t hear it while at 44.1KHz it’s in the audible range. Hardly anyone can actually hear above 20KHz (if even that much) anyway, which is why 44.1 was picked for the CD specification in the first place.

Where high sampling rates can sometimes pay off is some plugins can sound a little better at e.g. 96KHz because of how they’ve been programmed. It seems something to do with how they process high sampling rates means there are fewer audible artifacts once the audio is reduced back down to 48 or 44.1. Craig Anderton picked up on this a couple of years ago or thereabouts and there are some discussions about it on the old Cakewalk Sonar forum.

[Artcutech]

Thanks for that TLW, I’ll definitely be holding off on Catalina for the moment as well, will have to upgrade to 10 at that time, looking forward to see how 10 handles in my workflow. I think 96k sample rate would work great with these new t-racks tape plug-ins (I think I read somewhere that they are up sampled to 96k, that’s why they are so heavy on cpu and also the convolution) Ableton would completely take a crap if I switched to 96k haha and freezing and flattening is out of the question at the moment but might revisit as I can barely get by with a few instances spread around along with other heavy cpu-plug-ins in some of these huge arrangements, which was one of the reasons I started thinking about upgrading. In the future might look into getting a computer with fastest most powerful cores and not how many(although would help also) like you mentioned

[Artcutech]

Just wanted to ask how far off you think we are from from computers or Daws being able to spread every element of a Daw(Ableton specifically) across all cores as in not dedicating a single track to a specific core leading to overload? making sure the other cores pick up the slack and are maxed out before any cpu overload happens

[TLW]

A thought has just occurred to me. If you’ve not done so already download Intel’s Power Gadget. It can be very useful for showing what the cpu is actually doing.

On my 2.2GHz quad-core i7 MBP as soon as I reduce the audio buffer to 128 samples turbo mode kicks in and one core leaps straight up to maximum turbo speed and stays there. I assume that’s the core doing the audio driver stuff. Neither Live nor Apple’s own monitoring software indicates much in the way of cpu or audio system strain until I start using heavy duty plugins, nothing gets near overheating and everything works as I’d expect which is good, but I think it shows that maximum possible core speed can still matter.

[fishmonkey]

Just wanted to ask how far off you think we are from from computers or Daws being able to spread every element of a Daw(Ableton specifically) across all cores as in not dedicating a single track to a specific core leading to overload? making sure the other cores pick up the slack and are maxed out before any cpu overload happens

audio processing is very demanding, as it requires an extremely reliable continuous flow of computation to avoid glitches.

the thing is that we have multiple core machines now because the speed of a single core is largely limited by the physical technology. having multiple cores works pretty well when there are multiple independent tasks that need doing, however there are unavoidable overheads and synchronisation issues involved in splitting tasks up and spreading them across multiple cores.

also, many tasks cannot be readily split into chunks and spread across cores. if you think about how most audio effects work, future calculations are usually highly dependent on previous calculations. think about a reverb: the sound source is being bounced around in virtual space, and the generation of the reverb is an unfolding process. how would you split that across multiple cores?

[Lazzaku]

in my opionen it's just money thrown out of the window.
1) an 8 core Intel cpu is so fast that it will last the next decade when it comes to music production. i know times where quad-cores were the non-plus-ultra. and a lot of tracks were produced on pc/macs weaker than nowadays smartphones.
2) i don't want to know what an Intel 18 core will cost. as example, i paid 500 bucks for my i9 8 core.
3) it's not like a self build pc. These machines aren't build for upgradability. it's possible but you injure yourself badly - the psu is not covered, you could die.

[Artcutech]

Sup TLW, I haven’t downloaded the intel power gadget but will now that you mentioned it, thanks for the heads up

I appreciate the wealth of knowledge fishmonkey, Sure helps me understand the basics of a multi core setup and its limitations in dealing with audio processing specifically

Hi Lazzaku, after all that’s been said in this thread I think the upgrade would be throwing money away and am grateful for what my setup is able to do already which is quite a bit.

Those setups your speaking of that were not as powerful as cell phones, I use to have one and it was a joy to work on as most of everything was done on outboard gear and recorded into the daw and whatever mixing plug-ins were used were extremely forgiving on cpu, nowadays one newer high quality top teir plug-in on every track can really weigh your whole arrangement down quickly, some obviously are better than others as far as cpu, but I’m just saying, plug-ins in general have become so demanding

I have 22 universal audio cores which helps but i find myself only reaching for a couple of their plug-ins anymore, a 1073 preamp channel strip on every track being one, Hopefully we’ll see them come with a new wave of plugs or re-do their classics(with the latest modeling and sonic characteristics) to sound as amazing as what some of the other companies are coming with now and that will still be supported by the uad2 cores

[TLW]

I think with both PCs and Macs the time the cpu alone was the most crucial thing ended some years ago.

My first audio computer was an Amiga A1200 with an upgraded 68030 cpu, maths co-processor and factory hard drive. Which suddenly makes me feel very old...

At some point since then the balance of the whole system became more important than brute cpu power as bottle-necks appeared that were nothing to do with the cpu but to do with feeding it data fast enough; things like hard drive speed, how much data the drive controllers could shift, PCI bus speed, low latency audio drivers and, for DAWs, acceptably quiet cooling and mechanical noise levels.

One thing seems certain - no matter how powerful the computer is there will always eventually be software that pushes it to the limit. To complicate things DAWs have unusual requirements because they are an attempt to make non-real-time multi-tasking operating systems work as closely as possible to a real-time system - while running resource-intensive multi-tasking software. To complicate things even further system stability is vital in a DAW. No-one wants a crash in the middle of a gig or half-way through tracking a really great performance, especially if the performer is a paying client.

All of which makes for a set of requirements full of contradictions and potential problem areas. Maybe quantum computing, if it ever lives up to its potential, will be the next big leap forward. I guess we’ll have to wait and see.

[steko]

The Lounge's Multicore-Performance-Squad: working 24/7 for you to get to the bottom of things!

[Artcutech]

I feel what your saying TLW

i’m thinking the next computer I buy will be a quantum computer with a living core