u-he

Synth Master Jedi · Post by **Synth Master Jedi** » Tue Nov 19, 2013 11:09 pm

I downloaded this newest version and there seems to be a timing issue, at least on my system and DAW.

I am using Logic 9 and Snow Leopard.

Here are two short bounces of a cover track that I was doing. The first one is using the prior version of Diva and the second one is using the newest version of Diva. The bass and pad sounds are all Diva. It's easy to hear that the bass is tight in the first version but sounds kind of off in the second version. Besides swapping out the DIVA component, the two versions are identical.

Previous DIVA version
http://www.filedropper.com/divatest1

Newest DIVA version
http://www.filedropper.com/divatest2

Urs · Post by **Urs** » Wed Nov 20, 2013 2:14 pm

Arrrrgh. We had a most stupid bug, 4 lines of code before my first diagnostic step

MIDI Note timing Issues will be gone shortly

Adrian B · Post by **Adrian B** » Wed Nov 20, 2013 5:00 pm

The Multicore LED doesn't change colour with the LED colour control. Intentional?

When I load a factory patch, use the main window's browse patch left/right buttons, then switch the osc module to digital, the graphics (knob position and selected waveform) don't reflect what I hear until I make a change. I can give more precise steps to duplicate later if needed.

Diva VST-3, in 32-bit Cubase on Windows.

Incidentally... what does multi-core actually to? Does each voice get its own core or is it more sophisticated than that? Is DIVA compiled with extended CPU instructions (SSE)? Don't answer if it's proprietary knowledge

a

Urs · Post by **Urs** » Wed Nov 20, 2013 5:13 pm

Adrian B wrote:The Multicore LED doesn't change colour with the LED colour control. Intentional?

bug - fixed in next revision

When I load a factory patch, use the main window's browse patch left/right buttons, then switch the osc module to digital, the graphics (knob position and selected waveform) don't reflect what I hear until I make a change. I can give more precise steps to duplicate later if needed.

Diva VST-3, in 32-bit Cubase on Windows.

Sounds much like our struggle with VST3. We highly recommend using VST2 plug-in until we rewrite our VST3 layer (there's a conceptual discrepancy between VST3 and our framework which we hadn't caught until recently - we need to rewrite VST3 support for a more simple - and more predictable - subset of VST3).

Incidentally... what does multi-core actually to? Does each voice get its own core or is it more sophisticated than that? Is DIVA compiled with extended CPU instructions (SSE)? Don't answer if it's proprietary knowledge

In multithreaded mode Diva indeed creates a thread for each voice. From Sandy-Bridge CPUs on, the load is pretty evenly balanced over the cores. The overall CPu load is however not reflected in the performance meters of the host software - use Task Manager or Activity Monitor instead

Weasel-Boy · Post by **Weasel-Boy** » Wed Nov 20, 2013 8:56 pm

Urs wrote:Sounds much like our struggle with VST3. We highly recommend using VST2 plug-in until we rewrite our VST3 layer (there's a conceptual discrepancy between VST3 and our framework which we hadn't caught until recently - we need to rewrite VST3 support for a more simple - and more predictable - subset of VST3).

Urs: I'm just curious. Can you tell us what are a few advantages of the VST3 format over the VST2 format. Is there some real compelling reason to use the VST3 version over the VST2 version?

abique · Post by **abique** » Wed Nov 20, 2013 11:15 pm

Weasel-Boy wrote:Urs: I'm just curious. Can you tell us what are a few advantages of the VST3 format over the VST2 format. Is there some real compelling reason to use the VST3 version over the VST2 version?

Sometimes bugs can be considered as features

Urs · Post by **Urs** » Thu Nov 21, 2013 12:20 pm

Weasel-Boy wrote:
Urs wrote:Sounds much like our struggle with VST3. We highly recommend using VST2 plug-in until we rewrite our VST3 layer (there's a conceptual discrepancy between VST3 and our framework which we hadn't caught until recently - we need to rewrite VST3 support for a more simple - and more predictable - subset of VST3).
Urs: I'm just curious. Can you tell us what are a few advantages of the VST3 format over the VST2 format. Is there some real compelling reason to use the VST3 version over the VST2 version?

The main reason would be distribution over processes and physical machines. So you have the DAW and the Editor on one machine, and the DSP calculation on another.

This is also the main advantage of AUs and AAX. However, with AU and AAX this is easy to achieve for us. Whenever our editor needs to know something from the DSP part, it can say "Hey, I need that value" and the DSP part says "Ok, here it is". In VST3 however it's like "Hey, please send me that part..." <silence> "allright, I call you back shortly".

That means, in VST3 the communication is asynchronous. That makes distribution easy on the host side, but it makes some very simple and effective paradigms impossible to implement on the plug-in side.

The workaround for us is to give up distributability for VST3.

Another advantage of VST3 is that it's easily shelled, i.e. have multiple plug-ins in one binary file. I was of the impression that it's difficult to do in VST2, but I recently learned that it might be as easy. Thus, hmmm, not sure if that's a valid advantage then.

Yet another advantage of VST3 is that it's all Cocoa on Mac. No need to drag Carbon code along.

All other advantages that are, well, easily achieved in VST2 as well, but some need small add-ons to the API (timestamped parameter changes, side chains, automatic bypass on silenxe, scalable UI, anything)

zvenx · Post by **zvenx** » Thu Nov 21, 2013 1:21 pm

Urs wrote:
Weasel-Boy wrote:
Urs wrote:Sounds much like our struggle with VST3. We highly recommend using VST2 plug-in until we rewrite our VST3 layer (there's a conceptual discrepancy between VST3 and our framework which we hadn't caught until recently - we need to rewrite VST3 support for a more simple - and more predictable - subset of VST3).
Urs: I'm just curious. Can you tell us what are a few advantages of the VST3 format over the VST2 format. Is there some real compelling reason to use the VST3 version over the VST2 version?
The main reason would be distribution over processes and physical machines. So you have the DAW and the Editor on one machine, and the DSP calculation on another.

This is also the main advantage of AUs and AAX. However, with AU and AAX this is easy to achieve for us. Whenever our editor needs to know something from the DSP part, it can say "Hey, I need that value" and the DSP part says "Ok, here it is". In VST3 however it's like "Hey, please send me that part..." <silence> "allright, I call you back shortly".

That means, in VST3 the communication is asynchronous. That makes distribution easy on the host side, but it makes some very simple and effective paradigms impossible to implement on the plug-in side.

The workaround for us is to give up distributability for VST3.

Another advantage of VST3 is that it's easily shelled, i.e. have multiple plug-ins in one binary file. I was of the impression that it's difficult to do in VST2, but I recently learned that it might be as easy. Thus, hmmm, not sure if that's a valid advantage then.

Yet another advantage of VST3 is that it's all Cocoa on Mac. No need to drag Carbon code along.

All other advantages that are, well, easily achieved in VST2 as well, but some need small add-ons to the API (timestamped parameter changes, side chains, automatic bypass on silenxe, scalable UI, anything)

Re: the first one wont other developers have that problem too?

Rsp

Urs · Post by **Urs** » Thu Nov 21, 2013 1:44 pm

zvenx wrote:Re: the first one wont other developers have that problem too?

Rsp

Yes, probably.

However, there's a "simplified VST3" which I guess is what most are using now.

Or, if they use VSTGUI or JUCE, Steinberg ships the SDK with a solution for these.

Weasel-Boy · Post by **Weasel-Boy** » Thu Nov 21, 2013 9:05 pm

In VST3 however it's like "Hey, please send me that part..." <silence> "allright, I call you back shortly".

Sounds vaguely like what I got calling their support number

just kidding...

Thanks for the info, Urs. Sounds like a lot of under-the-hood stuff with no major user operational impact between the two versions. At least that's what I've observed using the two types in the same Project on Cubase 7.0.6.

DrGonzo · Post by **DrGonzo** » Fri Nov 22, 2013 7:30 pm

Hi guys - do you think there will be any further changes to the sound engine? What I'm basically asking is if it's safe enough to start making your own sounds with the beta?

/C

Urs · Post by **Urs** » Sat Nov 23, 2013 10:42 am

DrGonzo wrote:Hi guys - do you think there will be any further changes to the sound engine? What I'm basically asking is if it's safe enough to start making your own sounds with the beta?

/C

No further changes in sound - as far as I am aware.

analogisdigital · Post by **analogisdigital** » Mon Nov 25, 2013 10:47 am

Hello,

when should Diva2 be released?

whyterabbyt · Post by **whyterabbyt** » Mon Nov 25, 2013 10:50 am

analogisdigital wrote:Hello,

when should Diva2 be released?

Since they've not actually said anything about a Diva2 at any point, how about 'after the stuff they have said they're working on,' of which theres a fair bit.

analogisdigital · Post by **analogisdigital** » Mon Nov 25, 2013 11:13 am

whyterabbyt wrote:
analogisdigital wrote:Hello,

when should Diva2 be released?
Since they've not actually said anything about a Diva2 at any point, how about 'after the stuff they have said they're working on,' of which theres a fair bit.

right, I mean the new Diva 1.3.xx not 2 - the one we are testing of course.

u-he

KVR Audio

Diva 1.3.0b4 (r2033)