Access Virus & Virus TI community since 2002 Virus TI Infekted

Go Back   The Unofficial Access Virus & Virus TI Forum - since 2002 > General discussion > General discussion about music production

General discussion about music production Discussion concerning music production, composing, studio work, sequencing, software, etc.

Reply
 
Thread Tools Search this Thread Display Modes
  #31  
Old 28.07.2014, 07:44 AM
Berni's Avatar
Berni Berni is offline
Veteran
Aged Veteran
 
Join Date: 24.02.2009
Posts: 743
Default

What is this NERD FEST? Couldn't make it to Comic-con? Make some music you sad git's
Reply With Quote
  #32  
Old 28.07.2014, 10:25 AM
MBTC MBTC is offline
This forum member lives here
This forum member lives here
 
Join Date: 16.04.2010
Posts: 1,082
Default

Quote:
Originally Posted by Berni View Post
What is this NERD FEST? Couldn't make it to Comic-con? Make some music you sad git's
@berni: Contributing something of value again I see, don't you have some self-promoting to do or something? Run along now and don't get under the grown ups feet while they're busy

@grs:
I'll give that a try a little later. I basically did the same thing by enabling monitor on all 25 instrument tracks (which plays them simultaneously, and then you can just hold a note down), so I think I will hit the wall at 25 again but will report back. Cubase CPU meter is a little different than Ableton et al in that you have peak load and average load (peak is like maximum which doesn't really tell you a whole lot other than you're hitting the ceiling occasionally, average is more like the meter in other DAWs that really impacts what you hear). I'm probably at about 60% average which is way higher than I would run in practice without freezing tracks, but then again a good fat unison poly lead with fx and a long release from some synths can eat up maybe 25-30% by itself at times.

Also meant to ask, are you comparing both PCs using the same external audio interface or do they have distinct sound cards?
Reply With Quote
  #33  
Old 28.07.2014, 07:06 PM
TweakHead TweakHead is offline
Veteran
Veteran
 
Join Date: 16.07.2011
Posts: 573
Default

Quote:
Originally Posted by Berni View Post
What is this NERD FEST? Couldn't make it to Comic-con? Make some music you sad git's
@MBTC this actually made me laugh!

On another note, you're right about most plug-ins not profiting much from new instructions on the CPUs, but that's not as linear as you say: for instance, Diva has multi-thread support now, but that doesn't mean that any CPU can handle it. Same thing for most host software (daw), they can increase their performance if some instructions are found, but will perform as good as it can without them.

I do agree, however, that we're not seeing major increases in performance for quite a while now. I feel a good way to get better performance (and less money XD) is to get something like UAD dsp cards and plug-ins. One can manage insanely huge track counts that way without much stress to the system. That and hardware synths (expert sleepers can deliver sample accurate timing for analogue synths), the Virus ti (working properly) also helps a lot to dispense with cpu-heavy plug-ins (isn't that why it costs so much?). All of this and an SSD disk can boost performance much more then with a CPU+motherboard/chip change I think.
Reply With Quote
  #34  
Old 29.07.2014, 12:28 AM
MBTC MBTC is offline
This forum member lives here
This forum member lives here
 
Join Date: 16.04.2010
Posts: 1,082
Default

@Tweak: yes, I'm always interested in DSP related options. I'd love to see them come out with a PCI card that could run dedicated instances of Dune2. It seems once any device is placed outside the case, you just have one more cable component to introduce potential lag (USB, Thunderbolt, Firewire etc) even if in theory they shouldn't.

@grs: Well I never noticed it but it seems as though Cubase limits me to 24 instrument tracks. That's more than I would need in a project (which explains why I haven't hit the limit) but it seems odd for their flagship product (this is not Cubase LE or one of the baby editions). I guess to really push it I would have to use a much more resource intensive patch.

My 870 is handling the 24 instances of Spire with BA Octaver fine enough (one note held down for 2 bars looped), with total usage in the 50-65% range depending on what I set as my buffer size on the Saffire audio interface software. My CPU usage in task manager pretty much reflects the visual bar in Cubase, 50-65% or so. The part I'm confused about is how you were able to see what each instance of Spire was using using Task Manager (or is that an Ableton thing?) Maybe I'm misunderstanding what numbers you were looking at?

I should note in terms of audio quality, all of this sounds like hell but because of muddiness caused by duplicating the same patch too many times (not the type of crackling that comes when the CPU is strained). It does pop a little at the beginning of the first bar, but I think this is because of the phases of the OSCs all kicking off at once.

The other interesting thing is that the demo of Spire is swooshing the white noise for all of these instances at varying intervals, which has to be creating a certain amount of overhead in itself. You can see the monitor for the noise varying on all the different tracks which is weird to observe.
Reply With Quote
  #35  
Old 29.07.2014, 02:15 AM
grs's Avatar
grs grs is offline
Veteran
Veteran
 
Join Date: 15.11.2004
Location: Au
Posts: 558
Default

I used a division of the cpu by the instance as a figurative value.
So if you go backwards from 25, when do you not get the pop? and is the master mix not overloading?
Also my i7-950 is in another room without my Fireface so I will have to re-test to get the same buffer and card comparison. But the numbers seem right at the moment, maybe one more instance.
Reply With Quote
  #36  
Old 29.07.2014, 02:54 AM
MBTC MBTC is offline
This forum member lives here
This forum member lives here
 
Join Date: 16.04.2010
Posts: 1,082
Default

Well I discovered that the popping I was describing was mostly because of the copy protection white noise. So, to eliminate that and get all 24 instances into the same swoosh/silent cycle I saved and reloaded the project fresh, that way I can wait for the swoosh to subside before testing. Eliminates the pop entirely.

When you say is the master mix overloading, are you talking about the volume level on the stereo out master track (probably called something else in Ableton)? I brought the volume down because of course that would be a mess and clipping otherwise.

What is your overall CPU like on each machine when you run these tests? I'm at about 62% with a buffer size of 192 (I could reduce it down quite a bit with larger buffer size).

Also not sure if I mentioned but this is supposed to be a 2.93Ghz processor, I have it clocked to 3.1GHz, in my tests I was getting about a 10 cpu boost in DAW with that and it still runs cool (air cooled) so I just left it there.
Reply With Quote
  #37  
Old 29.07.2014, 03:22 AM
MBTC MBTC is offline
This forum member lives here
This forum member lives here
 
Join Date: 16.04.2010
Posts: 1,082
Default

Also just to be sure I understand your CPU calculations, you're saying 8 instances on the 4770k was about 24% total CPU, and about 64% on the 950, is that correct?
Reply With Quote
  #38  
Old 30.07.2014, 02:52 AM
grs's Avatar
grs grs is offline
Veteran
Veteran
 
Join Date: 15.11.2004
Location: Au
Posts: 558
Default

I think the only useful data to get from this sort of test is the amount of instances you can handle on a whole system.
I got what I needed in my own case to have a system I could do twice as much work on.
My old system had a different motherboard, windows 7, read emails, graphics and video editing apps etc. So all these per instance division are not real world reflections on that CPU's ability to make Soft Synth.
My new system is windows 8.1, SSD and music Apps only. So unless I put an SSD in my old system and do a fresh install I can't promise these figures to anyone.
Reply With Quote
  #39  
Old 23.08.2014, 08:27 PM
nutrinoland nutrinoland is offline
Pro
Pro
 
Join Date: 03.11.2009
Posts: 363
Default

Can it only use sine waves for Fm ? It has nice wavetables..would be great to be able to use them for FM
Reply With Quote
  #40  
Old 23.08.2014, 10:10 PM
MBTC MBTC is offline
This forum member lives here
This forum member lives here
 
Join Date: 16.04.2010
Posts: 1,082
Default

Quote:
Originally Posted by nutrinoland View Post
Can it only use sine waves for Fm ? It has nice wavetables..would be great to be able to use them for FM
Long answer first...

You have two "oscillator stacks" total (actually 3 but the third stack is more of a single sub osc than a stack, that doesn't have to be a sub osc, it can be a white noise or many other things but it does not have other complex features like FM, only the first two stacks do).

So, in each of those stacks you can have up to 32 voices (think about hypersaw detuning on the Virus and you get the idea). Those voices have a bunch of different options for changing how they detune (linear/Gaussian/random etc). If you set an osc stack type to FM, you now have a classic 3-operator FM panel for that stack (A-B-C) (and can change the algorithm, like A feedback into itself or A+B modulate C directly with B feeding back into itself, feedback adjustable on either of course).

So, figure in Unison possibility, (32 voices * 2 stacks * 8 unison * 3 operators) and you have 1536, or if you can live with only 768 operators on a single stack, you can use the other two stacks to use wavetables or virtual analog at the same time. (Actually this gives the illusion that you only have the granularity of 768 vs. 1536, but it's not like that at all, each of the 8 unison voices has its own stacks, own filter settings, etc so it can be as granular as you want).

I'm typing all that mostly to try to illustrate how it makes manipulation of so many FM operators so useful and easy to get great sounds. You're really only controlling 3 operators for any given stack and voice at one time, but that's still like controlling up to 24 separate sets of operator parameters if you need that kind of complexity. It also let's you work with multiple voices as if they are the same one if you want. That's part of the beauty of Dune 2 I think, other than just the sound. The way they workflow is designed. It's just so fast to work with.

Short answer next ...

Each of the A-B-C operators are fundamentally sine waves, but you end up with a range of sounds that's as good (better in many cases) than other FM synths and it's dead easy to get there. And a modulated sine isn't really a sine anymore
Reply With Quote
Reply


Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump


All times are GMT. The time now is 09:03 AM.
Powered by vBulletin® Version 3.6.4
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
Skin Designed by: Talk vBulletin
Copyright ©2002-2022, Infekted.org