I've been A/B'ing VA synth demos all week on my machine, a Mac Mini Server (2GHz Core i7). All I gotta say is Diva and ACE are phenomenal.

I'm looking to play these live, (i.e. I'm not really interested in playing in 'draft'), but my system just isn't powerful enough. Even at default quality settings many patches are totally unplayable (ex. single note max's out in Diva's 'BS THXish').

So I guess my question is, will the CPU performance get much better than what it is for either of these? Looking thru the posts it appears that might be the case. A rough guess would be that a 20% boost in CPU efficiency or the ability to multicore would definitely make this fully playable for me. Am I dreaming? Should I just give up, wait three years until I get a new system or what?

FTR, these synths are beautiful. While demo'ing a bunch of other synths, I turned off all the fx etc. for each so I could really just compare the raw oscillators and filters. None could even touch the Diva and ACE, both of which are not only more musical/organic to me, but they're also more responsive and hence eminently more playable.

Seems strange that you shouldn't get better mileage out of you Mac mini. Could you increase the buffer size in your host?

I know that a single instance of Diva can't be spread over more that one core.

THX patch is reputed to be one of the most CPU intensive patches.

I have Diva and can play it on my mid 2007 2.4 Ghz imac so it certainly seems strange this.

This worries me as I'm about to jump in and get a mac mini server to play these synths. Interestingly another user "aMused" that has recently acquired the upgraded Mac mini 2.7 Ghz i7 Duo core reports much better performance.

Could you outline your settings for playing these patches, nomber of voices, if stacked, audio buffer size and quality. it would be a great help.

Regards

Wade

We should have a considerable performance boost for Diva in a few weeks time.

Take into account that Diva's Draft mode is roughly the same quality as ACE's most accurate setting (until we add a similarly modeled filter to ACE, that is).

That said, we get quite a few voices out of our Mac Mini, it's kind of "okay-ish", but of course the THX one uses unison and thus uses multiple voices for a single note.

Cheers,

Urs

buy yourself a moog phatty,now thats a truly playable synth if ever i have heard one,no worrying about cpu bull either

I've been contemplating my next U-he purchase as well. There's a lot of buzz about diva, but I don't think anything has floored me the way ACE did when I first tried it. Maybe my musical tastes lean more towards unique sound design than super-accurate emulation...

Theres got to be something wrong there, I can play it in divine mode on a 2.5 Ghz C2D

Your i7 should wipe the floor with my machine.

Oh, yeah, duh, I forgot to change the buffer... thanks for reminding me. That helped a bit.

This is the BS THXish patch. And on a Mac Mini Server (2GHz Quad core, i7) it was maxing out.
I/O buffer = 256
4 voices
4 voice stack
mono
divine

Here's a screen grab:



However, if you put the buffer up to 512, it'll play, but only just. The regular mac mini is only a dual core, but it does run faster.

Oh well, apparently not.

Where the mini server kicks ass is when you need lots of tracks. But for stuff like this, it seems you're better off with fewer but faster cores, like what you have.

As I understand it (and I don't really), the VST/AU formats pretty much make it impossible for developers to utilize more than 1 core. Which kinda sucks... at least for me. Multicore is obviously the future, so I don't see how it's tenable for the AU format to remain the way it is for long.

You rock Urs! You're all over this forum. Most developers don't put nearly that much attention into their customer base. Says a lot about your company and your products.

I've been really getting as much info together about this multithreading and CPU usage issue. My idea is to get 2 or 3 or even 4 computers networked together. This way for the money to my way of thinking you will be able to get the most number of voices. if you used a single machine with 16 cores for example if the load doesn't get distributed effciently you essentially letting 15 cores stand idle.

But this multithreading IS possible in Logic or MainStage or Reaper.

Here the host takes care of the multithreading.

So the issue is then if each single core in the Mac mini at 2 Ghz is powerful enough for Diva with multiple stacked voice patches which are really CPU heavy as with THX.
This get's me rethinking my setup once again, and perhaps a 2.7 Ghz mac mini now, and the upgraded mac mini models when they come out.When this late summer...Actually the iMac 2.8GHZ i7 Ghz four core model is also looking very tempting suddenly. And it would look good along side of my old imac as well.

I'm really glad the OP pointed out the issue he's having with the mac mini server though..

I want 16-24 voices of Diva in Devine mode in my system. I also have Tassman/AAS which though an older synth it is very open ended and has instruments that easily take as much as the heaviest Diva patches. It sits on the shelf waiting for the system upgrade, after 1 year or so.

Heck even Surge/Vember is never really fully loaded in the patches. Which can uses 6 oscillators over to layers..

Razor by NI is also another huge CPU hog that I have bought and never used because of this.

Diva and ACE/Bazille are the only ones of these heavyweights that I would upgrade my computer setup for..

Finally it's also interesting to hear from Urs that they are working on making it more CPU efficient, though it's most likely a very difficult thing to implement. I spec my system for what the softsynths demands are Now...


Wade

4 voices stack is what's doing that - i can barely use 2 voices stack on my Q9650 3.0 ghz quad and that is at draft mode..

MMmmm I know I'm gonna get a lot of responses about this comment.... but here it goes.
First rather get a PC, I've got i7 930, with 6GB ram and can run whatever vst/vsti I want,
Secondly, get both synths, .... As far as I'm concerned Urs's products are of the best, if not the best. (but that's just my opinion)

In the interest of avoiding a thread a thread that ends with kittens ... perhaps if we start with baby sloths with stuffed animals

seriously - dial back on the voice stacking!! if i wanted 4 voice stacking i would need a new quad core... maybe you can do the stacking offline?