I'm using LASS loaded in Kontakt 5 inside of VE Pro 5 on a slave machine that is triggered by Cubase 6.5 (using VST3 64-bit). Anytime there is a tempo change in Cubase, the audio from VE Pro gets distorted. For example, if you have a "ramp" tempo change in Cubase, during the ramp, the audio is distorted, but at all other times, it is fine. Also, if you watch the VE Pro window during the tempo change, the focus randomly and quickly shifts between different Kontakt instances. In other words, the highlighted instances change rapidly like they do when I manually click them. Also, I verified that Kontakt is getting the tempo changes (shown on the master editor tab). However, it seems like the tempo changes are making VE Pro go crazy.

Here is my exact setup:

Cubase 6.5.0

VE Pro 5.1.11252

Kontakt 5.0.3.5812

Windows 7 64 bit

I've also tried Cubase 6.5.3 with the same problem. I just recently upgraded all of these tools, but I never experienced this problem in the last version of VE Pro 4. Thanks for the help.

I haven't found the exact cause, but when I use smaller Kontakt templates, the problem does not occur. I will experiment with different sizes to see if I can create a reproducible test. 

Good questions:

1) no

2) I'm using LASS 2, which requires K5

3) Small tempo jumps do not seem to cause a problem

Here is some other info I found. I currently have one VE Pro 5 instance for LASS2 with 5 Kontakt instances. As I start to delete Kontakt instances, the problem gets better (less distorted). If I get down to one instance, I *think* it is working correctly. This suggests a possible workaround of 5 separate LASS instances instead of 1 with 5 Kontakt instances. 

However, I would still like to figure out the problem because this prevents me from getting the most benefits from VST3. I also have a huge template on other machines, so I need to reduce VE Pro instances whenever possible. Thanks for the help.

OK, now we're getting somewhere. Could you try a similar thing with something like Stylus or Omnisphere, to see if those players can cope with tempo ramps? If they can. then it is definitely a Kontakt problem, LASS problem, or both. If they can't then the VEP programmers need to have a look at it.

The other thing to look at is whether or not things are better or worse if you change the multi processor settings in Kontakt (I can't remember whether or not you have done this).

DG

I just noticed that during the ramp tempo change, the CPU usage in VE Pro jumps to ~99%. This is definitely the problem, but I'm still not sure of the cause.

In my latest tests, the CPU during a tempo change is over 90% during a tempo change and only 20-30% at other times. Unfortunately, my previous "fixes" (either reducing the Kontakt instances or reducing the size of each instance) don't actually work because a single reduced instance still has the CPU around 85% during a tempo change. So, there actually isn't a workaround because even if I replicate VE Pro instances, the CPU will just jump back to 100%.

I've tried almost everything I can think of, with the exception of reverting back to VE Pro 4.

Sorry for the swarm of messages, but I keep finding new information. If I completely mute everything in Cubase, so there is no midi data being sent to VE Pro (or Kontakt), the CPU usage still jumps to ~100% on my VE PRO slave.during the tempo change.

I never thought about tempo sync, apparently I never found the need. however I looked at it, I have external sync enabled in every case; I have projects that double the tempo or more in Cubase. But, I am using Kontakt 4. This seems like an issue for NI per Kontakt 5 to me.

I too have issues with tempo changes using Cubase 6.5x and VE PRO.  I am not sure if it is just Kontakt or not, but my template is gigantic and I run a sound card buffer of 128.  I noticed that when mixing if I increase my sound card buffer to 512 I can mix the cues with no issues what so ever that I hear at 128 samples.  So I assume by me increasing to 512 that it is lowering the cpu usage across all machines getting rid of the problem.  In other words I think this shows more evidence that there is a cpu spike happening on the slave machines during tempo changes.

-Danny