Cherry Audio Forums

This subject has been raised before but here goes again with a feature request that to me seems blindingly obvious.

There needs to be the option to switch on some kind of performance meter in VM.

Yes one can use a general purpose CPU performance tool to look at CPU load but that's a bit like trying to find out what time of day it is by using a light meter.

Unless I've totally misunderstood how things work, what causes audio glitching is buffer underrun (also known as buffer underflow).

There are supply, demand and batch size aspects to this. Supply is the sum total effort of all the modules in a patch computing the audio data and filling up the buffer. This obviously varies according to the patch and the power of your computer. Demand is the need of VM to read the computed data at the sample rate. Then there is the batch size which is determined by the number of samples in the buffer. Generally the larger the buffer the more the system can cope with variations in supply (but at the cost of latency).

Distortion, clicking, popping and drop outs are all caused by the supply side of things not being able to fill up the buffer quickly enough for the demand side of things.

This is of crucial importance to almost everyone using VM yet we are almost completely in the dark as to how able our computer is to match supply with demand.

I realise that a meter that tells us how close we are to buffer underrun does itself reduce efficiency but we really do need the option of switching one on for the purpose of fine tuning our settings and trying to optimise patches that push our machine to the limit.

+1!

I know such meters from DAWs. It's always good to have CPU load under control for audio purposes.

hi,

I agree.

a) maybe also a posibility for sampling frequency of 22050 Hz?
b) copying (duplicating) modules or groups of modules using Ctrl+C?
c) the ability to slowly render complex patches without dropouts?

thanks and greets tom

This would be nice. Some modules draw more computer resources than others when inserted and also when running. It would be handy to have a meter that immediately shows the performance change (after a module has been inserted, for example) rather than to have to wait until a patch has been completed to hear the audio degradation.

VCV Rack has this feature on a per module basis. CPU draw is shown numerically (%) as well as graphically - the yellow overlay shows resting state CPU draw as well as spikes during usage. It's indispensable when you're trying to optimize a patch, since modules with duplicate functions (e.g. two different mixer modules) often use different amounts of CPU power.

+1 on the meter.

VCV Rack's metering is very sophisticated but I'd be happy if CA would just wake up and provide the most basic super crude application wide meter.

VM not providing any mechanism at all for monitoring performance (beyond - hey the audio is crackling) is becoming increasingly frustrating.

I'm actually considering buying the VCV Host so that I can run VM inside VCV Rack and at least that way be able to monitor VM's performance. Such a ridiculous state of affairs should not exist.

ColinP wrote: ↑Tue May 24, 2022 8:53 am

I'm actually considering buying the VCV Host so that I can run VM inside VCV Rack and at least that way be able to monitor VM's performance. Such a ridiculous state of affairs should not exist.

that*s pretty funny

xones2358 wrote: ↑Sat May 28, 2022 8:30 pm that*s pretty funny

Yeah, but not in a good way.

I'm evaluating VCV Rack at the moment and the VST Host add on may actually be worth the $30 as I really do need to be able to get more information about VM performance and it looks like CA have no plans to add even basic metrics anytime soon. I don't want to support a competitor to VM but needs must.

In case anyone is vaguely interested in my opinions, VCV Rack looks far more low-tech than I expected. Being able to look at module source code with a right click is an eye-opener. Yes there's no doubt plenty of naff code in VM but the use of Java enables lots of automatic optimizations, while in VCV Rack it looks like people are hand coding SIMD calls. Something I haven't done since the late 1990s!

Also although VCV Rack CPU load is OK I was staggered to see how much GPU it uses. On my Radeon equipped laptop it uses up over 50% of GPU capacity while doing almost nothing. In comparison it's hard to get anything above about 1% GPU use when using VM. Open GL is probably the culprit and VCV Rack is effectively locked into that, so things won't improve in the future as Open GL is pretty much abandonware.

Although VM looks far more polished and blows VCV Rack out of the water on GPU use, there are still some lessons CA could learn by looking at VCV Rack.

ColinP wrote: ↑Sun May 29, 2022 5:40 pm
Also although VCV Rack CPU load is OK I was staggered to see how much GPU it uses. On my Radeon equipped laptop it uses up over 50% of GPU capacity while doing almost nothing. In comparison it's hard to get anything above about 1% GPU use when using VM. Open GL is probably the culprit and VCV Rack is effectively locked into that, so things won't improve in the future as Open GL is pretty much abandonware.

One of VCV Rack's solutions to the heavy GPU draw is to offer variable frame rates. I'd be interested to know what GPU capacities you experience at the different frame rate settings.