Korg Wavestate and Opsix for Win/Mac

[SLiC]

I also think Reason Algorithm is the closest to Opsix, if I am completely honest I may even prefer the sounds I get from Algorithm a bit more so far and it was only 60 bucks when I picked it up…Opsix has a nicer GUI and in some ways is a more conventional FM synth than Algorithm which of course only runs as a VST in the Reason Rack…

[crickey13]

Algorithm which of course only runs as a VST in the Reason Rack…

Too bad about this, seems like a cool synth.

[Biome_Digital]

Tried the demos today. Here's my thoughts/opinion from a sound designer's view point.

Out of the 2, I prefer the GUI of the Opsix, For that reason, I didn't look any further with the Wavestate so can't comment on that.

The Opsix seems quite cool, but tbh, I would much prefer to use Serum, Dune, Pigments, Avenger, Hive, etc.. with better GUI's, the sound isn't any better than these synths either.

Now I can see it, I am going to have another look.

Also, they were glitching and going slow in Reaper, it made them unusable for me. Did anyone else have performance issues with in Reaper?

The price is high in my opinion.

Edited this after finding the "stupidly obvious" resize menu!!

[r10k]

Out of the 2, I prefer the GUI of the Opsix, although they are both too small, and I couldn't see a way to resize them.

Not to sound like a legend, but for each app it took me 5 seconds to find how to resize them. Interesting you couldn't find the menus...

[xalama qo]

Free Opsix patches shows how interesting it can sound.

That's the good stuff right there! Very nicely done. Aaah that reverb is so smooth

[PAK]

Fair enough. But opsix sure is doing something weird. Here's the CPU usage after a few minutes of startup but not playing any notes:

Yes, it's grinding away even if you aren't playing any voices. Like the Roland Cloud ACB stuff does...

Yep. So the two plugins generate CPU load for everyone, which doesn't bear much relation to the polyphony used, correct?

When you then leave the plugin to idle (no MIDI note input received) it will generate an even higher CPU load on some (all?) configurations. This would show as a varying percentage depending on what you're using to measure CPU (A 4 core CPU under Windows might display a constant CPU load of around 25% generated, for example).

My conclusion is either
1) The polyphony / CPU management is broken right now, and/or..
2) This stuff is running inside of an emulation layer.

Take yer pick. If it's the later then labelling them as "native" was particularly disingenuous on Korg's part.

I ran several tests on different Windows/DAW configurations (Reaper and multiple versions of Cubase), and all were the same. Any VST3 suspend settings, and even the denormal setting on Reaper, made zero difference. In fact, Reaper introduced an additional bug here where OpSix (And I'd assume WaveState, though I'd have to double check) generated pops and clicks until I disabled the VST3's GUI rendering. Cubase didn't suffer the issue.

[r10k]

Fair enough. But opsix sure is doing something weird. Here's the CPU usage after a few minutes of startup but not playing any notes:

Yes, it's grinding away even if you aren't playing any voices. Like the Roland Cloud ACB stuff does...

Yep. So the two plugins generate CPU load for everyone, which doesn't bear much relation to the polyphony used, correct?

When you then leave the plugin to idle (no MIDI note input received) it will generate an even higher CPU load on some (all?) configurations. This would show as a varying percentage depending on what you're using to measure CPU (A 4 core CPU under Windows might display a constant CPU load of around 25% generated, for example).

My conclusion is either
1) The polyphony / CPU management is broken right now, and/or..
2) This stuff is running inside of an emulation layer.

Take yer pick. If it's the later then labelling them as "native" was particularly disingenuous on Korg's part.

I ran several tests on different Windows/DAW configurations (Reaper and multiple versions of Cubase), and all were the same. Any VST3 suspend settings, and even the denormal setting on Reaper, made zero difference. In fact, Reaper introduced an additional bug here where OpSix (And I'd assume WaveState, though I'd have to double check) generated pops and clicks until I disabled the VST3's GUI rendering. Cubase didn't suffer the issue.

There's some chat over at VI Control about things on the Mac side. M1 Macs have lots of different things happening.

[PAK]

There's some chat over at VI Control about things on the Mac side. M1 Macs have lots of different things happening.

Yep, but I'm experiencing the exact same issues, as those on OSX/M1, except I'm running under both Windows 7 and Windows 10 (21H2) with Intel, in both Reaper (latest version) and Cubase (multiple versions). On Windows someone might have an 8 core system and think 12.5% CPU usage is reasonable, where it's actually pegging one of their cores at 100%..

[r10k]

There's some chat over at VI Control about things on the Mac side. M1 Macs have lots of different things happening.

Yep, but I'm experiencing the exact same issues, as those on OSX/M1, except I'm running under both Windows 7 and Windows 10 (21H2) with Intel, in both Reaper (latest version) and Cubase (multiple versions). On Windows someone might have an 8 core system and think 12.5% CPU usage is reasonable, where it's actually pegging one of their cores at 100%..

Haha for sure. What I mean is, not all M1 systems are seeing problems. I'm not. Things seem to be running fine. That said, I am an amateur when looking at this stuff

[SLiC]

Out of the 2, I prefer the GUI of the Opsix, although they are both too small, and I couldn't see a way to resize them.

3 dots on the right for all options- same as UHE plugs I think? 3rd item down in menu, resize

[PAK]

Haha for sure. What I mean is, not all M1 systems are seeing problems.

At least the very limited M1 configurations should make things easier to figure out there. It's just interesting that the exact same issue (increased CPU usage on idle) should be present on different platforms and architectures.

Best I can tell, the lack of polyphony management also impacts everyone (You'd expect 64 notes to create much higher CPU usage than 1 note, for example, but this doesn't quite appear to be what happens in either OpSix or WaveState). But the more powerful your CPU, the less prone you might be to notice (or care about) the issue. I just hope these are bugs and Korg haven't went down the (emulated) Roland Cloud ACB plugin path on this one..

[r10k]

Haha for sure. What I mean is, not all M1 systems are seeing problems.

At least the very limited M1 configurations should make things easier to figure out there. It's just interesting that the exact same issue (increased CPU usage on idle) should be present on different platforms and architectures.

Best I can tell, the lack of polyphony management also impacts everyone (You'd expect 64 notes to create much higher CPU usage than 1 note, for example, but this doesn't quite appear to be what happens in either OpSix or WaveState). But the more powerful your CPU, the less prone you might be to notice (or care about) the issue. I just hope these are bugs and Korg haven't went down the (emulated) Roland Cloud ACB plugin path on this one..

What does it mean if they have? I assume it's something they'll have to fix whatever way they've managed to do things...

[PAK]

What does it mean if they have?

It means that the amount of CPU usage will be similar whether you press 1 note or 20. This impacts how many other plugins you can run, and the more instances you have, the more it will impact things. Plus it's helping run up your power bill, because it's continuing to use your CPU resources even when it's sitting there doing nothing. Assuming they eventually managed these resources better than Roland, IF it's emulated it'd still use more CPU, for the same task, than if the plugin were truly running native code.

[beely]

I have reported the wavestate M1 performance issues to Korg and they have replicated it, so they are looking into it.

I've also let them know similar issues that I experience have occurred on Windows, and linked them to the relevant posts on this thread.

This is a demo of what I see in Logic, for instance:



In the cases of plugins with performance problems on Native M1, Logic's 100% core overload does not correspond with the Activity Meter thread load. This happens with both these Korg plugins, and Roland Cloud's ACB plugins too, which are unplayable here.

For "well-behaved" native plugins, eg Repro-5, the core activity *does* correspond with the Activity Monitor load. So there's something going on, and hopefully they can sort it out...

[r10k]

What does it mean if they have?

It means that the amount of CPU usage will be similar whether you press 1 note or 20. This impacts how many other plugins you can run, and the more instances you have, the more it will impact things. Plus it's helping run up your power bill, because it's continuing to use your CPU resources even when it's sitting there doing nothing. Assuming they eventually managed these resources better than Roland, IF it's emulated it'd still use more CPU, for the same task, than if the plugin were truly running native code.

Well, all that I knew, but thanks for confirming I do appreciate how the name isn't well chosen if there are things being emulated.