Ladies and Gents. I’ve been having some glitches that tell me it’s probably time to do a new build. My current system is an ASRock Desktop Mini with an Intel i7-8700 and 2 Samsung SSDs. I used an AudioBox USB96 for Audio and Midi. Put the whole thing in a small custom box with a fan. It has served me well.
So what are my fellow Cantabile users powering up with? RMI is out of my price range, and I’m kind of happy with the Mini-PC/NUC format.
Thanks in advance!
Rick
Hi Rick! I have a live performance machine very similar to yours, built about two years ago (i5 gen 12). Mine is in a 2sp rack case. Only a few notes, mostly from my son the IT guy.
I wanted mine to rack mount, but that is a personal use decision.
We went with a full sized motherboard. There were rumors that the ITX and smaller boards might be discontinued, but also because we could get an ATX power supply at the drop of a hat.
Look to get at least 32 GB RAM, especially on Windows
I think l’m running Win11 with no trouble, but l use all soft synths, not much is sample based. I’m using a Behringer UMC1820, l want to use more of the outputs for things (metronomes, etc). Go ahead and replace your drives, l’d say, but clone them, so everything you want is basically “in the same place”. Much of the emphasis now is on multi-cores, maybe look for one generation older on the CPU, they’ve had the bugs worked out, and you can start with a smaller (i5) CPU, and move up if you have to. Mine has worked reliably, my biggest issue early on was the USB addressing thing, which l’ve pretty much eliminated, now. Best of luck!
Hi Rick
Take a look at my latest gig rack guide. I recently upgraded after running an Intel NUC since 2017 to an ASUS NUC14RVKU7 Core Ultra 7
The key thing I learnt in this upgrade is that a lot of Windows performance tweaks no longer seem to work on modern processors/WIN11, necessitating a utility like Process Lasso (which I was using anyway) to do the hard work for you.
Hi Derek
I am wondering if Process Lasso might help me. As you may remember from earlier threads I ran into crackling problems, even after upgrading from an i5 to an i7 processor. I had to take to have all vsts off in each song, and only switching them on when needed, which meant a lot of re-jigging song states. I have since upped the RAM from 16G to 32G but still got the occasional crackle, which relates to Time Load. Advice form guys here was to check the buffer size, and I halved it. So far, so good, but I would have expected to have seen much better improvement from the RAM upgrade. It might be related to core useage, and if so I am lost as that is way beyond me!
So would Process Lasso be of use to me? And what level of tech knowledge do I need to use it?
Will do. I’m wondering if anyone has had issues with AMD platforms. Also, my current NUC has a desktop processor, but it looks as if most companies are moving towards the laptop versions.
Rick
No problem with AMD, my Gmktec K8 plus has been super stable, and handles anything I’ve thrown at it.
Hi, @Sausagefingers (David)
It is always hard to say, but absolutely no harm in trying, and you can try before you buy
Process Lasso has a lot of parameters you can play with, but the basics are pretty easy. In a task manager type view, right click over Cantabile and select Priority/Always/High to say that you want to run Cantabile at high priority, set the power plan to “bitsum highest performance”. You can of course do a lot more (and happy to share my settings) but that gets you started.
They are, and the main issue is core parking and stopping that (again where the old ways of doing this no longer seem to work), and also the fact that you now seem to have a mix of performance cores and efficiency cores. But Bitsum Process Lasso and also their other product ParkControl allows you to get on top of this
I have invested and straight away there is an improvement, but strangely I then got crackling from a couple of vsts it hasn’t happened on before! Effectively it had all 4 processor cores at maximum! But there is already an improvement. I will read the help info and see where I go from there. But actually I am teetering into computer territory I have never dared to even consider before, so I am on an interesting journey!
Tanks for the advice so far!
All up and running, following your basic set up advice, @Derek. Just a puzzle or two.
Historically I have been having problems with audio crackling, linked to the Time Load parameter. I doubled the RAM to 32G, but it remained, and I was advised that the Time Load issue might be to do with the buffer size, and the time needed to read the buffer overloading the processor. So I reduced the buffer size from 512 samples to 256 samples, and it did help, but I am still puzzled why the extra RAM seemed to have no bearing on it.
Having invested in Process Lasso I thought “Great! Here we go”, only to find not only audio crackles at times, but also CPU overload, a couple of times leading to a complete crash! However, instead of running into problems at anything over 100% Time Load, I was now seeing 110% and even 115% Time Load without a problem.
So I reverted to a 512 Samples buffer size, and everything seems better. Apart from another issue:
I loaded a set list with 27vsts, the highest vst load yet, and Memory Load on Process Lasso shows 75% to 80%, but after a while, this drops to about 56%, although I have no idea why! The songs in my set lists have all vsts turned off to start and finish, so there is little CPU load, but RAM useage remains at the maximum on both Cantabile and Process Lasso. So why does the load percentage drop (and then increase again) over time, regardless of any playing activity?
Does anyone have any suggestions?
First: reducing your buffer size will not reduce time load (and thus any “crackling” caused by buffer underrun) - quite the opposite. The larger your buffer is, the easier things are on your CPU, so typically when you run into CPU-related crackling issues, the solution can be to INCREASE your buffer size, not DECREASE it. But with 512 samples, your buffer sounds pretty high already - you should be able to work with that easily if you tweak your actual song / instrument setup. I run a 200 plugin pre-loaded setlist at 128 samples on an i5 CPU…
Second: RAM and CPU are two completely different elements of your system performance. Extra RAM means that you are able to load more instruments and samples into the fastest memory you have on your PC - that is required to access stuff instantly without having to load it from disk / SSD, which slows things down significantly. High time load means that your CPU is struggling to process your audio data through all the VST effects and to generate the audio signal from your VST instruments. That is mostly related to the complexity of the algorithms involved and the number and speed of your CPU cores that you can allocate to these algorithms. Adding memory (i.e. space for super-fast storage of data and samples) will not help your CPU when it’s struggling with the calculations.
That isn’t plausible - Time Load of over 100% means that your CPU takes longer to process the content of a buffer than the time allocated to play the current one - there will definitely be artifacts in that case - even if you don’t hear them…
Your memory load of 75-80% with 32G RAM suggests that you are using pretty massive sample libraries. Depending on the instruments you are using, there may be quite a high CPU load involved in initially loading these sample libaries into RAM - that may be contributing to keeping your CPU pretty busy without any notes playing at all. There may also be some re-allocation and clean-up of your sample data behind the scenes, leading to the reduction after initial population - another issue that can keep your system busy without actually playing any notes. Not knowing what your “27 vsts” are, I can only speculate, though, but I suspect you are straining your system with (too) many sample-based instruments. Maybe you can illustrate what these “27 vsts” are?
@Torsten has commented on everything I was going to say (and more) so I won’t repeat the Qs! 
I run on a buffer of 256 samples to be on the safe side as I have some pretty heavyweight VSTs - sometimes several in the same song, and I don’t feel that latency in the style of music I play.
Thanks, Torsten. That makes a lot of sense. I do know the difference between RAM and CPU, but not so much on how it all works. I spent 30 years away from music and computing, which is pretty much the history of MIDI. I can’t remember which version of Windows took us from DOS plus an overlay into Windows NT, but until then I was fairly ok with the digital realm, but it all left me behind, and it is a huge learning curve!
The 27 vsts is the largest set list I have ever had - I put it together, then was asked to add another couple of songs at the last minute, or I would have put it into two lists! However, my “crackling work round” was to have each song start and finish with everything switched off. The hope was that only the libraries I needed for each state in each song would be switched on, theoretically reducing demand. Of course that didn’t reduce how much was loaded into RAM, but did get rid of the problem.
One of the songs was particularly problematic. It had the Friedlander Violin split across the keyboard with Sonixinema’s Contemporary Solo Cello (which is a beautiful sound), running in Kontakt Player 7. Frieddlander is particularly CPU hungry, so I ran it in its least demanding form, but still cause Cantabile to crash. I was aware that Kontak 7 was known for doing odd things with RAM - stealing it and n ot letting it go, so made some change.s. I upgraded to Kontakt Player 8, and used Native Instrument’s Cremona Quartet Ensemble to replace both the other vsts. I actually uses more RAM, but is far less demanding on the system.
Just that one change has got rid of all the problems, including the varying Memory Load, and CPU load so far has been considerably reduced, even running the huge Set List!!!
So Bye Bye Kontakt Player 7, which hopefully was the culprit. It now means I have got to go through every song and change from Kontakt Player 7 to 8.
Once again, thanks to you guys for your untiring help. It is greatly appreciated!! 
I gotta say… I worked in IT for 20 years, but Torsten just has a way of explaining things!
So here is my take on Kontakt. My first foray into the VST word was with a gizmo called a Muse Receptor. It used a proprietary version of Linux, and Wine… a “compatibility layer” that allows you to run many Windows apps in Linux. Kontakt came along for the ride. The Receptor platform was powerful, but a real PITA because of the fact it was… proprietary.
When I looked to get away, I saw a little clip somewhere about Cantabile. Everything about this platform impressed. Brad’s dedication to providing users an amazing program, the community, you name it. And early on, I realized that Kontakt took up way more resources than many of the other “options.” So I started getting away from all of their products.
And here is another place that Torsten has come in handy. Search his posts! He has twiddled and tested quite a few VSTs!
And thank you all!
Rick