Is there any host sequencer on Windows that can effectively use 32 cores/64 logical cores?

[Amelia70]

Are there plugin synths on the market today that a 4GHZ processor can not cope with? Is this why single core speed is so important? I never had an issue with kontakt which is the sample engine all our libraries use, and it has it's own multithreading. Really, I am at a total loss as to why anyone thinks I would get better performance with a large 200 track project with an 8 core cpu vs a 32. I just flat out don't believe it if I'm to be completely honest.

FWIW, I've tried to explain why high clock-speed is important.

- Not every process in a DAW can be multi-threaded (spread across multiple cores).
A developer from FLS explained this in detail (months back).
- The lower the latency at which you're working, the more important single core performance.
ie: Monitoring in realtime thru software EFX/processing at a 32-sample ASIO buffer size is not something that lends itself to heavy multi-threading. You can verify this by looking at the load of each core.
- I'm not sure if this has changed recently, but UVI's Falcon wasn't able to use multiple cores.

If it comes down to choosing clock-speed or cores, clock-speed is the single most important factor.
Xeon CPUs often have *significantly* slower clock-speed (but significantly more cores).
If you've ever used one for DAW purposes, they're an expensive performance hit compared to a standard CPU (fewer cores but significantly higher clock-speed).
A CPU with 16 cores running at 3GHz will not outperform a CPU with 8-cores running at 5GHz.
At the risk of repeating, performance increase from adding cores doesn't scale 1:1.
IOW, Quadrupling the number of cores does not quadruple performance.

If you're running at higher latency settings... and using a DAW that has "Hybrid Buffering" (where tracks that are merely playing back are using a *large* playback buffer), that's going to make more efficient use of multiple cores. If you're working at the smallest ASIO buffer size (say 32-samples), that's not going to make efficient use of multiple cores.
You can see this for yourself with Cubase.
If your audio interface allows, set it to a 32-sample ASIO buffer size.
Load a project into Cubase... and note the Audio Performance meter.
Now, switch off the Activate Multi-Processing option... and note the Audio Performance meter.

I'll have some specific figures for the 3970x in the next couple of days.
If it can run all 32 cores at 4.5GHz, I'll be amazed... and the performance will be fantastic.
If it can't run anywhere close to all cores at 4.5GHz, it'll be underwhelming.
Details to follow...

Sorry but you said a 16 core at FOUR ghz won't beat an 8 core at 5. Now you are saying a 16 at 3 ghz and I agree with that too for sure. So we don't disagree after all.

I already explained we ONLY use kontakt when using VI, that's IT. The rest is insert effects and external stuff.. It doesn't matter what the general fastest is for most people, here I was asking about my personal way of working.

Well the low sample buffer thing must be a windows problem.. I know for a fact Logic perfectly multithreads across multiple cores at 32 buffer. I never needed to in PT cause of HDX but using HDX as an interface in logic without access to the DSP makes it an ordinary core audio interface. I am able to get extreme amounts of tracks even on this computer at 32 buffer in logic with those tracks armed for record and you can clearly see it spreads the load evenly. One thing I find curious is that Logic will only fill up what seems to be half the cores on it's performance meter for tracks that are monitor armed, and reserves the others for the high buffer for playback, but it works perfectly.
Maybe I am looking at this all wrong and should be getting a more powerful mac and just using Logic which is cheap and cheerful and I can sell all the HD interfaces and get a nice fast core audio interface.

I already SAID Cubase is problematic, I know this, I have been following what tuff cat says (pls is that his name? I Know it's something like that, forgive if wrong).
Cubase is broken wth multithreading with live armed tracks/asio guard disabled and the manufacturer doesn't care which means it's not suitable for me as I will need to stream large numbers of external inputs at a low buffer. TC says Cubase 10 can only use 6 cores for the low latency tracks, which will not be suitable for us and be a total waste of a high core count cpu. We only work at 64 in PT when playing Kontakt.. it has no bearing on the external monitoring as that's all being done via DSP. I could set it to the highest buffer, and the signals coming in would still be low latency at 48k. it's under 2ms at 48K.

If you like I can take a video of me arming 64 external inputs at 32 buffer in Logic, audio streaming into all of them simultaneously, and putting 2 effects on every track and also creating some reverb and delay aux/sends. Maybe you'll be amazed. I have no idea how PT behaves under Core audio as we use the DSP. Is there no windows DAW that can properly multithread at ultra low buffer? If that's the case then I'm not really interested, cause we need the cores for the eventual mix which will be hundreds of tracks with effects and as I said a heavy amount of IK tape. You have to try and not think in general terms but this specific workflow and if you still think a 5ghz 8 core machine will do what I need, we will continue to go around in circles as it will not, not unless I invest in a DSP solution again to aid.

I guess this is why I created the whole topic, which DAW can multithread properly.. I did mean when at low buffer with incoming signals. That's the most important aspect if I am to give up PT. Now I understand why you say high clock speed matters, cause I was going by what I was able to do in OSX on this old slow mac, and it's not an issue here, well certainly not with logic. I presume there would be windows DAWs that could do the same thing. I am starting to think windows might be the wrong way to go, but in any case, I will still try my friend's mac pro that she bought for her video studio when it arrives, but that's 2 weeks away unfortunately as there is a long ship time on them.

After reading this topic I tried the demo of studio one, and it is useless on MAC at low buffer, so maybe that is one where it's different on Windows? It could only input monitor about 1/4 of the tracks Logic could at the same buffer with same effects. Again I can't compare pro tools as that's using DSP. And yes, it makes sense, if the machine was a modern cpu running at a toasty 5ghz, Studio one would have coped much better. I understand that now.

BUT just to finish off once more.. once all these tracks are recorded to audio and the vi's the same, it becomes a mixing challenge with one of the most demanding plugins on the market over a huge number of tracks..Just to clarify, are you saying the 5ghz 8 core will reign supreme even THEN?
And if so, please explain how.. I am sorry you said you tried to explain but really you were explaining in a way that you expected the person to understand exactly what you meant and have sort of disregarded my own personal experiences with the computer in front of me right now, again and again.

Anyway, if this is all the case (what you say), then the answer to the entire topic is a resounding NO really, isn't it?

[Amelia70]

Ok i have got it.. I found a limited edition Intel processor that still has a few in stock overseas called the 9990X, it is expensive, the same price as the threadripper 32 core by the time I import it and pay tax. It is designed for all 14 cores to work full time at 5GHZ! Surely this would trounce the 9900K, and get me closer to where I need to be.. good amount of threads for playback mixing and high clock speed for monitoring at low buffer.. What do you think of that Jim?

Also Jim, I still do not understand why you are saying the threadripper will only be useful at 4.5ghz. Are you saying, that if all 32 cores are 4.0 ghz, not more, that 4GHZ isn't enough to handle playing kontakt at 32 or 64 buffer with a few effects? This is the bit that confuses me cause I can do that right now.

[Freaqpeak]

Now imagine how would it be if Intel had a 16-core 5 GHz monster. A 32-core Threadripper couldn't touch it, or it would be slightly above in certain multithreading workflows, but the difference in clock speed would prevail for the most part.

Imagine if TR was running all 5 GHz cores.
,.. I am 100% sure that the DAWs, games and apps in general will be better at multicore usage in the coming years. Win10 is already better since the Ryzen 3 release. Btw, I saw that some major game devs are now doing work to make games take advantage of multicores. IMHO Intel screwed us the customers for years now with their schemes, Intels only answer is a 50% (!!) pricedrop.. Now, that is some crap they have been pulling on us. But thats another topic.

5 GHz is more or less the highest we can go, and that has been true for many years. My current DAW, an older Intel i7 is running 4.6 GHz OC. I dont see Intel making any 6 GHz CPUs next year or the year after that.

So the future is going with more cores.

With that said, I probably still would have bought Intel again if they did indeed had the best CPUs in the sub 1000 USD range, for my needs (many instances of Diva). I was eyeing the 9900 but after I saw how powerful the TR 2950x was on a friends daw, I decided to wait half a year for results of the new Ryzen. In the end I waited for 10 months to buy my new system, since 3950x was delayed.. hope its worth it...

[stamp]

As I see it. If you are running a low track count with loads of effects go with an 8 core 5ghz cpu. But if you are running hundreds of tracks with just a couple of effects per track, the more cores the better even if you are running 500-600mhz less per core. Remember 1 track->1 core; thats how every daw distributes processing power.
In an ideal world you would buy a cpu with 32 cores at 5ghz , but unfortunatly such cpu doesn't exist, yet.

[stamp]

IMHO Intel screwed us the customers for years now with their schemes

So true.

[THE INTRANCER]

I'm still rocking on an Intel I7 920 quadcore 8 thread CPU at 3.7 Ghz, which is 10 years old tech wise, and a Focusright Pro 14 audio interface. I've been using computers since the 1980's and if there's one thing I've learned. It doesn't matter how fast your computer is, you'll always reach the limits in what it can do, sooner or later. The computer is only as good as the person that is using it and that's never been so apt than in the world of music production and music technology with computers. Using a DAW, isn't that far from being a mechanical engineer when building a car, or in this case producing a piece of music. The world of external hardware synthesizers have not vanished off the face of this earth yet... and you don't need a super computer to use them with a ridiculous amount of cores. Unless you're doing big 3D rendering projects, running a render farm... working on big video effect productions, for all but those who have an unlimited amount of money, it's really somewhat of an overkill. Unless you know how to optimise in the art of music production, then you will forever be unsatisfied with whatever computer you have...That's my two cents.. take it or leave it .

[tooneba]

I guess this is why I created the whole topic, which DAW can multithread properly.. I did mean when at low buffer with incoming signals. That's the most important aspect if I am to give up PT. Now I understand why you say high clock speed matters, cause I was going by what I was able to do in OSX on this old slow mac, and it's not an issue here, well certainly not with logic. I presume there would be windows DAWs that could do the same thing. I am starting to think windows might be the wrong way to go, but in any case, I will still try my friend's mac pro that she bought for her video studio when it arrives, but that's 2 weeks away unfortunately as there is a long ship time on them.

Have you tried to search and ask in Gearslutz too? There are also people running recording studio so I think you maybe able to get a hint.

[Amelia70]

You are telling me the 16 core at 4ghz will not beat the 8 core at 5ghz and I am willing to put money on that, that it will win both in VI polyphony and overall effects count by a large margin.

I am not telling you, the charts linked told you. 20 voices more and 26 instances more between 12-core 3900X and 8-core 9900K is NOT winning by a "large margin". 4 more cores on a 3950X won't double those differences, things don't scale 1:1 when adding more cores, as Jim already said a number of times!

Now imagine how would it be if Intel had a 16-core 5 GHz monster. A 32-core Threadripper couldn't touch it, or it would be slightly above in certain multithreading workflows, but the difference in clock speed would prevail for the most part.

Listen, I said the 3950 x annihilates it. And yes, 20 voices is more than 10% in this case and is a sound beating. Showing cores ultimately wins out with kontakt (what I use exclusively) vs speed. I would say the 16 core would be more around 30% for sure. That's a lot.

Once again, I said the SIXTEEN CORE 4ghz will beat the 8 core 5GHZ. You even quote me saying exactly that.

I just tested the heaviest falcon patches on my mac and I can play them, albeit they do fill up a core but i can still add a couple of effects..
Yes it's very heavy.. since the 5ghz is not DOUBLE per core of my computer, it means it won't be able to stack two instances of the flacon patch I just played.. which means it will only get one per core anyway. Once again, 16 cores wins.

Besides Reaper, no one has actually bothered to answer the original question, at all. Amazing, yet all anyone is trying to do is school me on what to buy and imagine they know my workflow. I said how many times now, that effects processing for playback tracks is my number one goal.. Using 3 or 4 kontakts per project in a sea of hardware means I need more CORES and a program that knows how to use them properly. There is nothing I haven't been able to play in kontakt with my CURRENT machine, and i keep saying that too.. I give up. Done. You win.

[Amelia70]

As I see it. If you are running a low track count with loads of effects go with an 8 core 5ghz cpu. But if you are running hundreds of tracks with just a couple of effects per track, the more cores the better even if you are running 500-600mhz less per core. Remember 1 track->1 core; thats how every daw distributes processing power.
In an ideal world you would buy a cpu with 32 cores at 5ghz , but unfortunatly such cpu doesn't exist, yet.

Exactly. Thank you. Now I just need to find the right DAW that can distribute across a lot of cores.

[Amelia70]

And yes, there are virtual instruments that can eat a lot of CPU for breakfast. Falcon (it does NOT multithread, hence single core speed is paramount!), Omnisphere, Repro-5, Reaktor Blocks once you start adding stuff, Softube Modular, and then on FX side Acustica Audio, for example. But there are many more examples.

I don't use any of those and never will. I only have falcon installed cause it was free with pro tools.

[Jim Roseberry]

Well the low sample buffer thing must be a windows problem.. I know for a fact Logic perfectly multithreads across multiple cores at 32 buffer.

I'm not surprised by this at all...

It's not a Windows (or OSX) issue... let me explain

Logic uses "Hybrid Buffering".
Tracks that are merely playing back are processed using a **large** buffer.
Even when you've got your ASIO buffer size set to 32-samples, the tracks that are just playing back are processed with that much larger buffer. That's exactly why you see it making good use of multiple CPU cores... even when you're running at a 32-sample ASIO buffer size.
Tracks that are being recorded/monitored in realtime thru software; those are the only tracks that would be subject to being processed with the 32-sample buffer size.
The load of that would not scale well across multiple cores.
The reason why is pretty simple.
There's latency involved in spreading the load across cores.
With a 32-sample ASIO buffer size, you're talking 0.75ms (44.1k) to process the next audio buffer.

On the Intel side, if you're comparing the socket 1151 i9-9900k to the socket 2066 i9 CPUs:
We've already established that the 9900k can run all 8 cores at 5GHz.
The socket 2066 i9 CPUs have a max turbo frequency of 4.5Ghz.
The 10-core 9900x can usually run all cores at 4.5Ghz.
That's two more cores than the 9900k, but 500MHz slower clock-speed on each core.
We know that performance increase for adding cores doesn't scale 1:1.
The two additional cores on the 9900x doesn't increase performance by 20%.
Performance wise, this results in a "wash" for many purposes.
Obviously socket 2066 has quad-channel RAM and more PCIe lanes.
To best the 9900k for most purposes, you have to get into the 14-core 9940x.
The 9940x most likely won't run all 14 cores at 4.4GHz (typically it's around 4.1 to 4.2GHz).
The 18-core 9980xe won't run all cores rock-solid at 4.4GHz (not under heavy loads).
The 9980xe ends up similar to the 9840x (clock-speed wise) across all cores.
Yes, the 9980xe out-performs the 9900k for many purposes... but it's not as dramatic as the price difference would have you believe.

"Hybrid Buffering" is a genius way to stretch much more from a given CPU.
Essentially, you're running everything other than tracks currently being monitored (thru software) at high buffer settings.
Assuming the clock speed difference isn't radical (between the 9900k and whatever mega-core CPU you want to compare), an app using Hybrid Buffering would definitely benefit from those extra cores. If the clock-speed of the mega-core CPU is significantly slower, you end up with a wash... or even a performance hit.

A word about benchmarks.
If you've read about the Ryzen 3970x and the wonderful general-purpose benchmark scores, note that most of those tests are video or encoding related.
That's Ryzen's wheel-house (heavily multi-threaded scenarios)... and yes, it will excel.

Have the build underway... and realized the 360mm water-cooler block won't work with the adapter included with the 3970x. The 3970x includes a circular mount (fits NZXT and Thermaltake coolers). Won't work with Corsair Pro series.

[tooneba]

Speaking of intel 9900XE 9990XE, its TDP is 255W, you need large custom water cooling unless car engine like noise isn't the problem.

[Edit]

[THE INTRANCER]

Besides Reaper, no one has actually bothered to answer the original question, at all.

I'm actually surprised people have been so patient with you... and going by your reaction to those who have tried to help you.. They are unlikely to again due to your outright argumentative replies to them delivered by large walls of text. Your opening post is a huge wall of text which could have been split into direct point questions that would answer what you needed to know. Allow people to answer and be receptive rather than be confrontational to what information they provide. They don't owe you anything, but what they have given is their time... so be appreciative of that. If you don't agree, say nothing, that saves you time and everyone else in reading through what you have to say. Don't forget that there may be a lot of what you say, which people don't agree with too, so it works both ways.

[jonljacobi]

I think there aren't a whole lot of people doing what you do or with the hardware you're considering so they don't have the answer (including myself), but this is KVR after all, so everyone chimes in anyway. Pay it no never mind.

You might check forums (composers? studios?) where your scenario could be more common. I've seen some stuff on YouTube on large projects and on Ryzen as well. If you really want to be bold, ask the developers.

If I had to guess, I'd say that any DAW that can reasonably expect to be deployed in a professional studio can, or will in the near future be able to utilize that many cores.

[EvilDragon]

Remember 1 track->1 core; thats how every daw distributes processing power.

It's not always like that. As soon as you use busses, all tracks sending to a bus are processed on a single thread, because you have a serial processing dependancy then, which cannot be parallelized (at least to my knowledge).