@Nick Bellagio First of all: which CPU Model (Name) and Cooler are you using?

I have a 10 Core i9 10850K running macOS Catalina on a Z490 Vision G Board (G stands for Gaming, btw!). I tested the performance with a Logic benchmark project with around 190 Instrument Tracks at 32 samples latency and 5 plugins an each track (1 physical modelling synth plus 4 audio fx each), which makes for… well a LOT of Plugins.

Anyway, my point is: if you run into CPU issues with an i9 CPU you are either doing something wrong or you are having a serious hardware insufficiencies or configuration issue.

 

First thing you should try is without any OC.
So set everything in your bios to Default and check if its better.

 

'komplete audio 6 interface'
I had one of these once, things improved massively when I switched to an Apollo Twin.

 

This might be a good idea for total newbies, but you can (and on certain systems you should) definitely go beyond XMP profiles. For example, for Zen2 AMD processors, the optimal memory speed is 3733Mhz because of the Infinity Fabric. If you happen to have quality memory, you can easily overclock it and reduce latencies. AMD systems especially benefit from memory optimization.

 

It applies to anyone not aware. Novice and veteran alike.

As you said, "quality memory"...This is where the QVL come in. Memory tested by the board maker. You always take your chances with off-the-shelf memory not on the QVL. Little glitches to major crashing or no booting issues are common problems when a board & CPU can't register memory properly.

 

Have you tried the same projects in Bitwig?

 

I do this within a batch: https://pastebin.com/ndR2G4J7

 

Fantastic - that's very practical - thanks for the information.

 

This wasn't exactly what I meant. Obviously if you find your RAM in your motherboard's QVL, that's a good thing. What I meant that you should also consider investing in a quality memory, for example something that has Samsung B-Die chip in it. Overclocking will be a bit safer.

BTW, qualified vendor list doesn't always include all the compatible parts. My memory was not included in my motherboard's QVL. I asked the support staff if it would still work anyway, and they said yes, there should be no problem. For some reason this information has still not been updated on their site. Indeed, it has been as stable as it can be, and I have had no problem overclocking it.

 

that's the beauty of using Windows

i9 cores cannot do magically far better than i3 cores, just because you have more cores doesn't make things problem-free, single-core stack is what limits most low-latency audio stability, and on top of that most likely it's dragged down by your audio interface, unless you're using something like RME HDSPe AIO,

if you did some CPU overclocking, make sure your all cores are boosting to same frequency, and can sustain that frequency through entire workload, no jumping or spiking, and also make sure your RAM is being clocked accordingly to support the stability of entire system,
also you will probably get better results by disabling hyperthreading in BIOS completely as well,

when working with lots of virtual instruments, just freeze/bounce the tracks so you save LOTS of cpu power and prevent crackles

 

I was suddenly having serious cpu issues during recording one day. Fixed it when I found out that I'd actually changed my laptop out of PERFORMANCE mode into ENTERTAINMENT mode. Annoying and very much not entertaining.

 

One instance of Kontakt can have up to 64 .nki (I think it's 64!) in a Multi, using all MIDI channels, all going to separate Kontakt outs via Kontakt's mixer, and then represented as separate audio channels and MIDI channels in your daw.
This is better for cpu than multiple Kontakt instances.
Plus, why this is so true:

Is because daws still love working with audio more than lots and lots of MIDI.
The MIDI info of all daw automation also uses a lot of cpu.

Just as we should always consider using the old with the new in all aspects of life, within our field here, it is obvious this truth applies to using both analogue and digital, as well as the approach and subsequent philosophy of recording to a medium such as tape where we commited ourselves.
With the power we now have in a/the box, (which I was dreaming of in the 80s at least in regards of MIDI sequencing, synths, and sampling!), I know it's hard to not want to just push our machine to the limits because we can!
And because it allows us to not have to commit for as long as possible!
Are we going forwards, or going back?

You have a computer that is far better than mine to the order of a power of magnitude.
But so far, it's possible that you have only achieved what I would have with my computer if I had worked the way that you have been doing!
Around 20 years ago we had 1GB of RAM and I can't even remember the Ghz of the cpu!
So we were bouncing down all the time, but could still always run many tracks of audio.
And being much younger and still knowing of the proper studios, we wanted to be as authentic as possible.
It's not that we can abuse the power we now have, but possibly we can allow it to abuse ourselves in some way.
Which of course is actually a choice that one made, not 'allowed', and probably subconsciously because it was invisible having have been given to us.
It's deep.

Essentially, music is none of this.
And we can use this technology that was offered to us, (by-products of the hidden industries of war technology, surveillance tech, nasa tech, that leaks down to programming tech (films, 'news' programming)), and use it however we choose.
Not in the way it is implied we should use it due to the power it now has.
For instance, it helps people to see music more as a commodity, rather than what it truly is.
I don't think that's a good thing for the soul.

Btw, the NeoDymium magnets that EV were the first to use in both a CD horn flare for the original DeltaMax cab horn, and also in their 757 and 575 mics was nasa material.
And that's what influenced the Beta 58 from Shure.
Plus, the Ferro-fluid that was also used in the same CD horn flare of the Deltamax!? nasa material.
For more interesting stories closer to the source in regards time, look up Thorn and susequent Thorn-EMI; Seimens; etc.
Yet having alluded to all that, I am constantly thankful that Georg Neumann carried on forward and had success after the terrible bombing of Berlin, and that the Philips family managed to escape the netherlands when they did!
Hypthothetical now I guess, but it's possible things could have been very different for us if things happened differently!
And just so there is no misunderstanding here, I hope we can all see just how strong the bud I just piped was?! lol. oh dear.

edited: bad spells and newb grammar!
One paragraph even makes sense now!

 

Last edited: May 20, 2021

Modules with Samsung B-Die are on most QVL. Some boards don't work with Samsung modules (won't boot/erratic errors).
Some of the top rated Crucial & G-Skill modules cause constant reboot while training or, unexplained random crashes (instability) with some board/CPU combinations (AMD or Intel).

Further...Using XMP Profiles usually give the best performance for stable operation no matter what memory is used.

In my experience with Workstation (Creator) or Workstation/Server boards is that they tend to be stable. Most have a limited BIOS setup to ensure no system settings could be set incorrectly causing stability or performance issues. They also tend to have better memory options such as low powered & ECC (buffered & unbuffered) modules. The other feature is usually better separation of PCIe lanes for USB & SATA.

 

That's a great idea! But how would I exactly use your script?

I have a lower mid-range laptop as my primary device, and it's absolutely essential for me to use different buffer sizes for recording and mixing! A buffer size with a playable latency would never be enough in my case for heavier mixing loads.

This is a very important point. If you have many Kontakt instances, their CPU and memory usage stacks up quite fast. Also, if you have two instances of the same Kontakt instrument, loading them up in two instances of Kontakt would consume twice the memory, whereas loading both on a single instance of Kontakt uses only a single instance's amount of memory, and both instances share the sample pool.

 

It makes sense now that I've been reading about, I feel like such a dumbass for not having this occur to me before. I have a different buffer size on each driver I switch to. For mixing I use Asio4All but for recording live I switch to my interface driver.

But after learning all this I realized I could have just switched the buffer size for each task without switching drivers while also switching buffer sizes at different stages of my project to adjust the load it takes.

Seems to make perfect sense to me now why I always had to switch drivers or I would get latency when I would try to record live even if the load wasn't all that bad.

 

copy it to an emtpy .txt file and rename to .bat.
Still edit it with notepad3 or similar editor for your like.

Code:

:: Start batch if bluescreen. devices get's reactivated
SCHTASKS /change /TN "Lenovo batch start" /enable

This is special and u need to add a Windows task in Taskscheduler.
So if bluescreen the task is activated and runs on next start another .bat with activating the devices and disables the task theme self.
If all is fine after DAW exit the init .bat deactivates the device recover task with

Code:

SCHTASKS /change /TN "Lenovo batch start" /disable

In the kill_apps.bat is someting like that

Code:

:: stop Windows service
psservice.exe stop MozillaMaintenance

:: kills an executable
pskill.exe TPFanControl.exe

Hope it helps a little.. not really complicated. If someone have a hint to make it better tell me/us plz!

 

There's fantastic tool that you can apply to see which app of drivers causes a lot of trouble in your system. You can dl it for free at: h**ps://w*w.resplendence.com/latencymon_using. Some of the problems could being caused by network drivers like the ndis.sys. This tool did a great job for me to discover the culprit of a bad performing system!! Just a suggestion...already mentioned by poli.

 

Well see this is what im saying i feel like it should be able to handle a lot. I have an Intel core i9 9900k 8C/16T Overclocked, and DEEPCOOL castle 240EX watercooling. (Also) Thats a lot of stuff but where you playing them all at the same time, because I can have a lot of instruments loaded its when I try and play them I start having issues. Id say on average I probably have 3 instances of kontak (5 instances within each), plus a maybe 10 instances of massive, and maybe 20 audio tracks at certain points layered and a drum rack and it cant handle it. *Also I should mention Ive been randomly getting blue screen its not that frequent but it has been happening usually when im producing but thats pretty much all I do on my computer.

 

Last edited: May 18, 2021

Can you tell us which error message the BlueScreen shows?
Can you post the event here?
Windows + r -> eventvr.exe or directly by right-clicking on the Windows symbol -> Event Viewer.

There are two articles to read:
- Optimizing the audio settings for KOMPLETE AUDIO 1/2/6
- Native Instruments - Windows optimization tips for audio processing

 

This pc has an automatic overclocking. I have a komplete audio 6 interface. How does the interface play into all of this?