AVA Vocal Flow maxes CPU

[TheBassBagwhan]

I've touched on this before elsewhere, but now we're talking my 32C upgrade. Just one instance of AVA Vocal Flow blows my CPU to 99% up from around 30-40% tops. 
Any ideas?

Windows 10, i5 CPU, 8GB ram, SSD drives (yes, it's a bit low-spec, but still...)

[johne53]

I've got a pretty powerful system so it's not as bad as what your seeing - but it's still pretty bad!! 

A small session with no plugins uses 3% CPU and gives me a DSP reading of 4% - but if I insert just one AVA Vocal Flow plugin, CPU usage increases to 10% and Mixbus's DSP figure shoots up to 53% !!!

[TheBassBagwhan]

I've got a pretty powerful system so it's not as bad as what your seeing - but it's still pretty bad!! 

A small session with no plugins uses 3% CPU and gives me a DSP reading of 4% - but if I insert just one AVA Vocal Flow plugin, CPU usage increases to 10% and Mixbus's DSP figure shoots up to 53% !!!

So adding maybe 40% of CPU is on a par with me. Glad to know its not just me and my system. Makes the plugin unusable really...

Thanks for the info.

[johne53]

Here's something interesting... over on the General Discussions forum, someone's started a thread about a product called Audiogridder which allows you to transfer some of Mkixbus's processing power to a different computer - so I decided to try it. I had a few initial problems setting up Harrison's VST3 plugins but eventually managed it. So I loaded the same session as last time and checked that AVA Vocal Flow was still giving me 53% (which it was). I then tried the same plugin via Audiogridder and Mixbus's DSP reading came down to 16%.

One slightly strange anomaly is that if I bypass any plugin (that's being managed by Audiogridder) I see quite a steep rise in MB's DSP figure, rather than a fall which is what I'd expect. In normal use though, it does seem to improve things (on my system, anyway...)

[Dingo]

I've got a pretty powerful system so it's not as bad as what your seeing - but it's still pretty bad!! 

A small session with no plugins uses 3% CPU and gives me a DSP reading of 4% - but if I insert just one AVA Vocal Flow plugin, CPU usage increases to 10% and Mixbus's DSP figure shoots up to 53% !!!

I get something similar on my older system, a late 2012 (2.3GHz Quad Core i7 16GB RAM) Mac which is running 10.15.7. 
After the initial increase to about 50% the CPU drops after a minute or two and sits stable at about 30%.
That figure does not change when I add another 22 instances of VF and 3 of BF.
The session itself is about 30 tracks all with multiple sources / regions.
Does the CPU settle down after a while ? Does adding more instances increase the CPU ?
Is the higher CPU actually giving you problems ?

[TheBassBagwhan]

I've got a pretty powerful system so it's not as bad as what your seeing - but it's still pretty bad!! 

A small session with no plugins uses 3% CPU and gives me a DSP reading of 4% - but if I insert just one AVA Vocal Flow plugin, CPU usage increases to 10% and Mixbus's DSP figure shoots up to 53% !!!

I get something similar on my older system, a late 2012 (2.3GHz Quad Core i7 16GB RAM) Mac which is running 10.15.7. 
After the initial increase to about 50% the CPU drops after a minute or two and sits stable at about 30%.
That figure does not change when I add another 22 instances of VF and 3 of BF.
The session itself is about 30 tracks all with multiple sources / regions.
Does the CPU settle down after a while ? Does adding more instances increase the CPU ?
Is the higher CPU actually giving you problems ?

I'm putting together a quick project now, and I'll try giving the AVA time to settle down...

[johne53]

I get something similar on my older system, a late 2012 (2.3GHz Quad Core i7 16GB RAM) Mac which is running 10.15.7. 
After the initial increase to about 50% the CPU drops after a minute or two and sits stable at about 30%.

[...]

Does adding more instances increase the CPU ?

That's not quite what I see on Windows. I haven't tried multiple instances yet but what happens here with a single plugin is that the reading stays above 50% as long as there's audio on that particular track. But as soon as the audio clip ends, it drops to under 10% (or just 4% if I use the Jack backend). so I'd imagine that if the multiple instances are on tracks which don't have any audio, they won't add significantly to the DSP reading. The other problem with Vocal flow (here) is that I see regular xruns if I use it with my PortAudio backend - but I'm not seeing them either with Jack 

[TheBassBagwhan]

I get something similar on my older system, a late 2012 (2.3GHz Quad Core i7 16GB RAM) Mac which is running 10.15.7. 
After the initial increase to about 50% the CPU drops after a minute or two and sits stable at about 30%.

[...]

Does adding more instances increase the CPU ?

That's not quite what I see on Windows. I haven't tried multiple instances yet but what happens here with a single plugin is that the reading stays above 50% as long as there's audio on that particular track. But as soon as the audio clip ends, it drops to under 10% (or just 4% if I use the Jack backend). so I'd imagine that if the multiple instances are on tracks which don't have any audio, they won't add significantly to the DSP reading. The other problem with Vocal flow (here) is that I see regular xruns if I use it with my PortAudio backend - but I'm not seeing them either with Jack 

Today I added one instance of AVA on a vocal track and Mixbus 32 clipped at 100% CPU continuously. However, Windows Task manager measured MB as using around 55-60%.

[johne53]

Today I added one instance of AVA on a vocal track and Mixbus 32 clipped at 100% CPU continuously. However, Windows Task manager measured MB as using around 55-60%.

Was that with Audiogridder - or just using the plugin normally? Audiogridder's definitely worth trying if you've got 2 computers sharing a network.

[PBuryk]

I spent some time reading the latest version of the Mixbus 32C manual last night.
Please check out what it has to say regarding Buffer sizes with regards to CPU Utilization.
In short, it explains that a larger buffer expands the window of processing ability and
that the CPU utilization meter represents the percent of that window used to complete the
amount of processing during that moment of time - NOT the overall load on real CPU processing availability.
This suggests that a very powerful CPU will still exhibit the issues as described in this thread (as in john53's case).
Further, it explains that xruns should (or may be) exhibited around the time(s) where the CPU meter reaches 100%.
This all suggests that if one were to increase the Buffer size then then should observe a reduction of CPU meter and xruns.

I'm not at my studio at the moment so I can't verify this, but wanted to report it.

Cheers!
Patrick