Audio dropouts when rendering/freezing LinuxSampler tracks

[lemmib]

Hi!
Anyone here with experiences with LinuxSampler and GIG-files?

I'd like to freeze or render a track that contains
a) LinuxSampler (playing .gig-files) as Insert FX and
b) a Midi-Item with the notes the sampled instrument will play.

During normal playback everything seems fine.
However, when i freeze/render the track, sustained notes stop randomly before the note ends. When i unfreeze the track and freeze it again, the stops/dropouts occur at different positions (i.e. the bug is not reproducible.

Is this a known issue?
If not: What's more likely? Is this a bug in Reaper or in LinuxSampler?

I'm using Windows 7 (64bit), LinuxSampler 2.1.0.svn6 and Reaper 5.972.


Best regards and thanks in advance,
Lennart

[lemmib]

Update:
If i check "Limit apply FX/render stems to realtime" (under Preferences / Audio / Rendering), no dropouts occur when freezing tracks.
(Btw: Is it possible to mute the audio when freezing in realtime?)

This solves my problem, but of course it's slow as hell compared to non-realtime renders.
So, a proper solution would be still very much appreciated! :-)

[Thunderclash]

Hi Lennart!

I'm having the exact same issue, but with Kontakt. Sustained notes gets cut off randomly when you freeze tracks. It happens on all tracks and even if you try freezing them one at a time. The CPU isn't overloaded or anything. And I've tried with really high buffer sizes, still the same problem. Realtime rendering solves it though.

If anyone can shed some light on this issue that would be greatly appreciated.
The freeze function seems buggy in alot of situations unless used with realtime rendering.

Cheers!

Update:
I managed to isolate the problem in my case. The problem was Kontakt's DFD function, how much it reads from disk versus RAM, was set too small. I increased it and now things are working better. Even though I had the samples on a SSD, the disk couldn't keep up at lower settings.

[lemmib]

Originally Posted by Thunderclash

Update:
I managed to isolate the problem in my case. The problem was Kontakt's DFD function, how much it reads from disk versus RAM, was set too small. I increased it and now things are working better. Even though I had the samples on a SSD, the disk couldn't keep up at lower settings.

Thank you for your reply!
That seems to be the problem in my case as well, since LinuxSampler uses a similar efficient streaming technique (just did some quick google-research).

I will look further into this when i'm back home from work today and post an update.

Cheers!