Page 2 of 2

Posted: Mon Aug 27, 2007 3:50 pm
by Cochise
garyb wrote:who cares what the exact problem is.
you want a stable system. i'm sorry to say this, but i'd start over. from your description, there's a 50/50 chance it's the os itself. also, there really is no reason for RAID if you're only doing audio with less than 40 tracks......

It's not perfectly clear to me the 50/50 thing, however yours is a wise advice: I don't need RAID, and also I don't need all those disks inside my cabinet. It's useless I go reinstall with RAID again; I should had to do that backup sooner or later...



bassdude wrote:Nuendo 3.0.0.2 is not the latest version. Did this version use to work or is this the first time you are using Nuendo?
Yeah... thx I was almost forgetting about it. I was using Nuendo 3.0.02 before installing the Raptor, running the OS from the RAID 0 stripe, and I had a similar problem, but much less marked and much less often. It was happening something like one time each 40-50 min for a duration of 2-300 ms. Now it happens each 10 min about for a duration of 4-500 ms.
It actually may sounds more like a dirty/damaged audio CD having a brief loop anywhere (but faster than what I usually hear with CDs).

And that's not a nice thing while playing :D

Posted: Mon Aug 27, 2007 8:58 pm
by BingoTheClowno
Cochise wrote:

I had some try creating a registry key for boosting IRQ8 (cmos clock) priority, but was useless.
I'll have a last try disabling indexing service.
Right in that section there is key called Win32PrioritySeparation that you can tweak according to this rule:
The bitmask is divided into three, two-bit pairs (AABBCC). Each bit-pair corresponds to certain behavior. I've broken it down below:

The highest bits (AABBCC) specify whether the processor interval (the amount of time allocated to the threads of a process) is relatively short or long. The "middle" bits (AABBCC) specify whether the interval is fixed or variable. The lowest bits (AABBCC) determine the ratio of foreground threads to background threads. Certain values cause different behavior, depending on whether you are using Professional or Server. Here's a table that breaks down each bit-pair.

Value Meaning
Highest Bits Interval
00 or 11 Shorter Intervals (Windows 2000/XP Professional); Longer Intervals (Windows 2000/2002 Server)
01 Longer Intervals
10 Shorter Intervals
Middle Bits Fixed or variable
00 or 11 Variable length (Pro); Fixed length (Server)
01 Variable Length
10 Fixed Length
Lowest Bits Ratio of foreground to background threads
00 Equal and fixed. This value also overrides the Fixed/Variable value to Fixed.
01 2:1. Foreground processes receive twice the processor time of background processes.
10 or 11 3:1. Foreground processes receive three times the processor time of background processes
References:
http://blogs.msdn.com/embedded/archive/ ... 43141.aspx

Read this Win 2003 server optimization guide,

http://www.redbooks.ibm.com/redpapers/pdfs/redp3943.pdf

you can apply many of the tweak to WinXP.

Disable services (if you have ATI keyboard poler service running, disable that sucker). If you don't use your PC for Internet access, you can disable many related services.

Posted: Fri Aug 31, 2007 12:03 am
by TimingLess
Hi Cochise,

I have a RAID 5 SYSTEM with three Harddisks and while reading your post, I recognized that we have the similar problem.

Maybe the RAID Drivers are take to much.

Does anyone know, if it is a big big problem to make an Image and go back to
three Harddisks without RAID and playback the image.


Thanks a lot

DT

Posted: Fri Aug 31, 2007 1:26 am
by valis
These programs have run fine on Xp systems for years.

Some time ago VSTi's are not properly load balanced in Steinberg apps until v3 or v4 (forget which, might be v4) because they run in a separate process. I ran Nuendo since 1.3 and they fucked this up in 1.53 or so when Vsti was added. I even paid for Sx1 thinking it would be fixed, but back then it even yielded crashes on a dual Xeon, both with Scope *and with the RME Hammerfall that they sold rebranded as their "Nuendo" audio interface*. SMP mode was always buggier.

I moved to Logic and found that I enjoyed other things about it as well, besides the fact that I got better audio performance from it than I did from Sx/Nuendo with or without SMP mode (this is something I know to have been presumably fixed in v4, the 'all the time' processing of everything even silence). I would expect the VSTi load balancing to have been addressed by now though, considering the number of users who bleat about it in a positive way on Steinberg's forums.

Basically try disabling it and see if your stability increases. In fact I'd recommend this for the guy complaining about crackles. The downside is that Xp's threading, Sx/Nuendo's threading should be working properly so if that 'fixes' things there's still something else stealing cycles somewhere. Malware?