KVR Audio

schnapsglas · Post by **schnapsglas** » Sat Feb 02, 2013 7:40 pm

Sampleconstruct wrote:I started a moderated group for Diversion sounds on Soundcloud, check it out here:
http://soundcloud.com/groups/sounds-of-diversion

Hey, that sounds good. Dmitry seems to be a top guy (obsessed with guitar sounds and DSP perhaps even more than me) and Diversion... is now my go-to place for sound design unless it's sample based. Nothing else comes close to me.

sches · Post by **sches** » Sun Feb 03, 2013 8:28 am

Thanks guys for your kind words! I'm checking the crash Simon. Excellent sounds as always.

Martin Simon · Post by **Martin Simon** » Sun Feb 03, 2013 2:35 pm

Hi Dmitry,

I'm demoing the Diversion and I really like the synth.
I was just wondering...Does Diversion support multi-core CPU?
Can we expect more optimalisation for CPU-usage?
It brings my system down to its knees very quickly.
I was also wondering if the is any difference in quality when you render a track or realtime performance....In other words is the oversampling rate the same for rendering as in realtime?
Thanks in advance

ZenPunkHippy · Post by **ZenPunkHippy** » Sun Feb 03, 2013 2:58 pm

It is a bit of a CPU monster, but the sound quality is amazing!

One possible bug I've noticed is that changing the oversampling cause the Logic CPU metre to spike severely. Because the oversampling setting is saved with each preset, changing presets is causing regular CPU spikes.

I have not tested the Mac/VST or Win version yet.

Peace,
Andy.

sches · Post by **sches** » Sun Feb 03, 2013 3:55 pm

Hi Martin, the plugin is single-threaded. There are too many dependencies between voices and no algorithms which can be straightly split up into threads. So Diversion won't be multi-threaded. Also real-time playing and rendering have the same settings, no difference here. Diversion has already optimized, there will be no further performance improvements. The synth's architecture is heavy on CPU by design, not by lack of optimizations. The only way is to change architecture in order to reduce CPU load.

Thanks Andy, I'll check the spike in Logic. It shouldn't be severe, something around 5-10% CPU on my PC.

Sampleconstruct · Post by **Sampleconstruct** » Sun Feb 03, 2013 4:38 pm

Something dreamy with 2 patches:
http://soundcloud.com/sampleconstruct/d ... -diversion

and something edgy with 1 patch:
http://soundcloud.com/sampleconstruct/t ... -diversion

Martin Simon · Post by **Martin Simon** » Mon Feb 04, 2013 11:59 am

Thanks for your response Dmitry.

Actually, I found out that the thing which stresses my CPU the most is the AMP release parameter.

When I turn the amp decay and sustain up to max nothing really happens CPU wise...maybe increase of about 5%. But when the release is turned up it turns haywire.
Is this the way it should work? I'm working with Live 8.3.4(32 bit)
Thanks

Sampleconstruct · Post by **Sampleconstruct** » Mon Feb 04, 2013 12:20 pm

Martin Simon wrote:Thanks for your response Dmitry.

Actually, I found out that the thing which stresses my CPU the most is the AMP release parameter.

When I turn the amp decay and sustain up to max nothing really happens CPU wise...maybe increase of about 5%. But when the release is turned up it turns haywire.
Is this the way it should work? I'm working with Live 8.3.4(32 bit)
Thanks

Chiming in here - higher release->more voices used as each voice is still active until it has fully decayed, so when e.g. playing a 4 note chord and then another one, 8 voices will be active until the first chord has fully decayed.

Martin Simon · Post by **Martin Simon** » Mon Feb 04, 2013 1:02 pm

Ah yes,
But isn't that happening when the decay is turned up as well?
Obviously, this gets worse with high voice count, unison and rapidly played notes.

I'm a bit whining about this CPU thing, I know.
I guess I really like the synth, but it should be usable in a production, right?

sches · Post by **sches** » Mon Feb 04, 2013 1:33 pm

Sure Martin, number of voices played depends on AmpEnv release, while decay doesn't affect voice management. The longer release the more voices are playing simultaneously. This is very important for unison. High unison count plus long release is a real CPU killer. Try to keep release and unison count as lower as possible, otherwise DAW freeze function is the only way to deal.

BDeep · Post by **BDeep** » Mon Feb 04, 2013 2:11 pm

No worries Dmitry, in a few years CPU's will be even faster and it won't be as much of an issue. Look at Massive, when it came out, people thought it was heavy on the CPU, but now, it's considered quite light.

Sampleconstruct · Post by **Sampleconstruct** » Mon Feb 04, 2013 4:33 pm

I'm still having the "vanishing sound" problem, especially with temposyned sequences, sound drops out completely, then one has to flip through some other patches several times until sound is retrieved, then it drops out again and so on.

Sampleconstruct · Post by **Sampleconstruct** » Mon Feb 04, 2013 5:16 pm

3 new patches in one Urban SciFi Fusion Calypso whatever demo:

http://soundcloud.com/sampleconstruct/u ... usion-demo

biomechanoid · Post by **biomechanoid** » Mon Feb 04, 2013 5:44 pm

Sampleconstruct wrote:I'm still having the "vanishing sound" problem, especially with temposyned sequences, sound drops out completely, then one has to flip through some other patches several times until sound is retrieved, then it drops out again and so on.

yeah, I've noticed that happening too, but not found a definite way to reproduce it yet, altho it seems to happen more with some patches than others.

sches · Post by **sches** » Mon Feb 04, 2013 6:52 pm

I also got it few times and working on it. Something strange here, soon will fix that.

Diversion for MacOS released