Welcome Guest! To enable all features please Login or Register.

Notification

Icon
Error

Forum Jump  
Not too many VIs on the latest Mac Pro...
Last post Fri, Dec 12 2008 by temasterson_1870, 13 replies.
Options
Go to last post
Posted on Mon, Dec 08 2008 23:19
by 9u10m
Joined on Thu, Dec 04 2008, Posts 83

Hi all,

I've just bought SE, and tested it on a recent rig:

2x2.8GHz Quad core Xeon / 14Gb RAM / OSX 10.5.5 / FireFace 800 / Logic 8.0.2 / latest versions of VE and VI installed.

I can easily open 32 VIs and load them with universal matrices. But as I play each of them, I start hearing jitter when I reach the 22nd instance or so, then the sound quickly becomes inaudible.

During this process, the last CPU core rises progressively until it reaches a peak that won't go down, even if I stop playing.

The size of the samples loaded doesn't seem to make a difference.

I remember the first benches that I read about, with around 24 VIs instantiated on Pentium 4 computers, which leads me to think something's not ok...

I'm aware of the VE standalone through IAC workaround, but at the moment I can't bring the audio back into Logic once I've routed VE's outputs to IAC busses. I'd like to avoid using Soundflower if possible (any detailed info on that matter will receive a warm welcome...)


Thanks for your help
Posted on Tue, Dec 09 2008 00:18
by bobulusbillman
Joined on Mon, Jul 09 2007, Posts 80
I have the same Mac Pro; I've noticed this happens BUT only when your building up your template. I don't know if you've tried but once you loaded up all the VIs save your project then restart the computer and load up the project again. You should find all these problems will have gone.

I find the performance on the Mac Pro with VIs to be only hindered by the 32bit RAM limitations - in terms of processing I've really yet to hit a limit - you just have to know the quirks of the system!
Posted on Tue, Dec 09 2008 02:48
by jroadrage_1768
Joined on Fri, Mar 14 2003, Posts 50
I'm on a similar Macpro using DP with fairly high latency but I have yet to run into any real big CPU concerns. I have 4 templates in VE that I use simultaneously (2 standalone and 2 plugs, about 5 GB total) and it takes A LOT before I start getting dropouts. What are your latency settings? Are you streaming from a separate drive(s) for samples?

John Rodriguez
John Rodriguez
Posted on Tue, Dec 09 2008 02:57
by jroadrage_1768
Joined on Fri, Mar 14 2003, Posts 50
Also, I usually use an ADAT loop to route audio from VE2 standalone back into DP, but I have used soundflower and found it to be pretty reliable.
John Rodriguez
Posted on Tue, Dec 09 2008 12:01
by 9u10m
Joined on Thu, Dec 04 2008, Posts 83

Latency is 512 (x2 on VE server), samples are stored on an internal SATA hard drive.

The jitter also occurs after a fresh system boot.

I understand Soundflower isn't mandatory to get the audio back into Logic with a Fireface, as its inputs and ouputs can be toyed with in the soundcard's matrix. So if I can avoid using yet another software layer, I will... But then I guess I don't get the ticks inside the matrix right (anyone with a screenshot?)

About the ADAT loop part, I have 2 i/o's on my FF800, and no combination I try gives me any result...

Unless there is a small glitch that I can't put my finger on somewhere in my setup, I do feel like an utter idiot right now (unpleasant feeling)...


Thanks for your time.
Posted on Tue, Dec 09 2008 14:55
by DG
Joined on Wed, May 12 2004, Posts 8608

I don't have a Mac Pro, but my machine is very similar. Obviously using Windows I can work a different way, but at the moment I am using anything up to 100 Instruments (usually not more than 60 at a time though) without any breakup. This is at a buffer of 128. for more dense orchestration I have to raise the buffer to 256 for a clean, pop-free playback. If you are getting much less than this then your system doesn't seem to be working as well as it should.


Let me know if you want me to test anything on my system.


If this has been of no help, then just ignore me. As someone who used to earn their living from conducting, I'm used to it.  Sad


DG

Nuendo 6.03, 4.3
2 x Intel Xeon x5675 3.07GHz Hex Core
48GB RAM
Windows 7 (x64)Pro
RME Multiface II
Intensity
ATI HD5400 series graphics card
Posted on Tue, Dec 09 2008 16:16
by aplanchard
Joined on Tue, Feb 06 2007, New York City, Posts 144

Hey DG,

Interesting.  What are your system specs?

Posted on Tue, Dec 09 2008 18:16
by 9u10m
Joined on Thu, Dec 04 2008, Posts 83

I gather you're on a 64bit copy of Windows then, otherwise RAM allocating issues would simply prevent you from loading that many samples (unless you open discrete standalone instances of VI?).

If that's the case, then it could mean the problem comes from the 32bit limitations on our side. So hopefully this will all be water under the bridge in a few months.

By the way, the renegade core does calm down if I hit stop and get back to zero after having the CPU meter peak while playing several VIs in a row. I can repeat this endlessly. Now that to me looks like a bug...

I haven't had time to give another try to the audio looping issue concerning the MIDI through IAC workaround, but I'd still appreciate someone sending me a screenshot of their Fireface matrix setup if they can provide.


Many thanks
Posted on Tue, Dec 09 2008 18:49
by DG
Joined on Wed, May 12 2004, Posts 8608
aplanchard wrote:

Hey DG,

Interesting.  What are your system specs?

Nuendo 4.2.2, 3.2
2 x Intel Xeon 2.83GHz Quad Core
32GB RAM
XP64 Pro Sp2
RME Multiface II
Intensity
ATI HD3470 graphics card

DG

Nuendo 6.03, 4.3
2 x Intel Xeon x5675 3.07GHz Hex Core
48GB RAM
Windows 7 (x64)Pro
RME Multiface II
Intensity
ATI HD5400 series graphics card
Posted on Tue, Dec 09 2008 18:53
by DG
Joined on Wed, May 12 2004, Posts 8608
9u10m wrote:

I gather you're on a 64bit copy of Windows then, otherwise RAM allocating issues would simply prevent you from loading that many samples (unless you open discrete standalone instances of VI?).

If that's the case, then it could mean the problem comes from the 32bit limitations on our side. So hopefully this will all be water under the bridge in a few months.

What is the memory usage in Vienna when these problems happen? Anything over 3GB would be asking for trouble.document.write( '<'+'a'+' '+'h'+'r'+'e'+'f'+'='+'"'+'&'+'#'+'1'+'0'+'9'+';'+'&'+'#'+'9'+'7'+';'+'&'+'#'+'1'+'0'+'5'+';'+'&'+'#'+'1'+'0'+'8'+';'+ '&'+'#'+'1'+'1'+'6'+';'+'&'+'#'+'1'+'1'+'1'+';'+'&'+'#'+'5'+'8'+';'+'&'+'#'+'3'+'7'+';'+'&'+'#'+'5'+'4'+';'+'&'+'#'+'6'+'8'+';'+ '&'+'#'+'3'+'7'+';'+'&'+'#'+'5'+'2'+';'+'&'+'#'+'4'+'8'+';'+'&'+'#'+'3'+'7'+';'+'&'+'#'+'5'+'4'+';'+'&'+'#'+'5'+'3'+';'+'&'+'#'+ '3'+'7'+';'+'&'+'#'+'5'+'5'+';'+'&'+'#'+'5'+'1'+';'+'&'+'#'+'3'+'7'+';'+'&'+'#'+'5'+'5'+';'+'&'+'#'+'5'+'2'+';'+'&'+'#'+'3'+'7'+ ';'+'&'+'#'+'5'+'5'+';'+'&'+'#'+'5'+'0'+';'+'&'+'#'+'3'+'7'+';'+'&'+'#'+'5'+'4'+';'+'&'+'#'+'7'+'0'+';'+'&'+'#'+'3'+'7'+';'+'&'+ '#'+'5'+'0'+';'+'&'+'#'+'6'+'9'+';'+'&'+'#'+'3'+'7'+';'+'&'+'#'+'5'+'4'+';'+'&'+'#'+'5'+'1'+';'+'&'+'#'+'3'+'7'+';'+'&'+'#'+'5'+ '4'+';'+'&'+'#'+'7'+'0'+';'+'&'+'#'+'3'+'7'+';'+'&'+'#'+'5'+'0'+';'+'&'+'#'+'6'+'9'+';'+'&'+'#'+'3'+'7'+';'+'&'+'#'+'5'+'5'+';'+ '&'+'#'+'5'+'3'+';'+'&'+'#'+'3'+'7'+';'+'&'+'#'+'5'+'4'+';'+'&'+'#'+'6'+'6'+';'+'"'+'>'+'&'+'#'+'1'+'0'+'9'+';'+'&'+'#'+'6'+'4'+ ';'+'&'+'#'+'1'+'0'+'1'+';'+'&'+'#'+'1'+'1'+'5'+';'+'&'+'#'+'1'+'1'+'6'+';'+'&'+'#'+'1'+'1'+'4'+';'+'&'+'#'+'1'+'1'+'1'+';'+'&'+ '#'+'4'+'6'+';'+'&'+'#'+'9'+'9'+';'+'&'+'#'+'1'+'1'+'1'+';'+'&'+'#'+'4'+'6'+';'+'&'+'#'+'1'+'1'+'7'+';'+'&'+'#'+'1'+'0'+'7'+';'+ '<'+'/'+'a'+'>'); ">

9u10m wrote:
 By the way, the renegade core does calm down if I hit stop and get back to zero after having the CPU meter peak while playing several VIs in a row. I can repeat this endlessly. Now that to me looks like a bug...

Many thanks

AFAIK Logic allocates one core for each VI instance. If you are using Vienna Instruments this should not be a problem, but if you are using Vienna Ensemble it may be that you are just trying to do too much for only one core. Do things improve if you increase the number of instances?


DG

Nuendo 6.03, 4.3
2 x Intel Xeon x5675 3.07GHz Hex Core
48GB RAM
Windows 7 (x64)Pro
RME Multiface II
Intensity
ATI HD5400 series graphics card
Posted on Tue, Dec 09 2008 20:13
by 9u10m
Joined on Thu, Dec 04 2008, Posts 83

Thanks for your quick answer DG. Memory's just fine when the problem occurs, and everything gets back to normal if and when I hit stop and rewind, until I start playing a number of VIs again. This a one track / one VI template, I'm not using VE plug-ins.

I'd be interested in a feedback from someone in the VSL team on this CPU peak thing.


Thanks again
Posted on Thu, Dec 11 2008 08:07
by chriskard
Joined on Thu, Jan 09 2003, Vienna, Posts 206
Hello 9u10m,

Can you please send us your template where this happens? (support [at] vsl [dot] co [dot] at) 22 Instruments were doable on a G5 2.5 Ghz.
On a MacPro 3.0 Ghz we had 80 Instances playing without any glitches.

Btw, I just had the RME Madi card for a long time but no FF. I think you can use the FF Matrix OR the Loopback option to route the Audio back into Logic.
Have you tried to keep the Matrix empty and only use just an Adat cable?
I assume that the analog ins and outs are numbered from 1 to 8 and the Adat ins/outs start at channel 9.

Now set the Driver to FF in VE Prefs and in Logic, set the output from an instrument in VE to out 9-10.
In Logic´s mixer you create an Aux channel (cmd+alt+n) and set the input of this Aux to 9-10.
Voilà, now your audio signal from the instrument in VE is routed to the Logic mixer.

Attention!! Be sure to "kill" the IAC-Bus midi ins in the "Clicks and Ports" Environment of Logic.
IAC-Buses are ins and outs at the same time (No specific ports for midi in or out are available for the IAC´s)

This is a much more CPU gentle option to handle the audio compared to Soundflower or Jack OSX.

I hope we can solve this quickly.

best,
christian
________________________
Composer
VIENNA SYMPHONIC LIBRARY
Posted on Fri, Dec 12 2008 20:13
by temasterson_1870
Joined on Tue, Mar 25 2003, Posts 19
DG wrote:

AFAIK Logic allocates one core for each VI instance. If you are using Vienna Instruments this should not be a problem,

 

DG



This is not my experience with VI in Logic 8 / MacPro 8x2.8ghz. If I am ONLY playing back already pre-recorded MIDI the Mac distributes the load evenly across all 8 cores. But as soon as I select any of my VI tracks to record a new section, the Mac immediately shifts all the load to one processor, which usually causes peaks.
You cannot post new threads in this forum.
You cannot reply to threads in this forum.
You cannot delete your posts in this forum.
You cannot edit your posts in this forum.
You cannot create polls in this forum.
You cannot vote in polls in this forum.