I am using the latest build of Cantabile and running into performance issues with Omnisphere.
When I run Omnisphere with the same patches and settings within Cubase, I do not have the same issues.
In Cantabile, the patches are crackling at times with the monitor showing around %150.
I am running on an HP laptop with 16 GB RAM, 5000 RPM Hard Drive, Core I7.
Our next live show is this Saturday, so trying to get the system better optimized. I’ve played around with various settings withing Cantabile itself without much of an effect.
Set multicore to “Automatic”. This is recommended (physical cores only). Also have you tried set list pre-load? I don’t know which Omnisphere patches you run but if they are streaming from a 5000 rpm disk that could be the problem.
Except that the issue doesn’t happen in Cubase, according to the OP.
The only major difference in hosting would be how the CPU is handled.
I would see what happens if you turn multicore support off completely and let Omnisphere manage itself.
Sorry you’re having this problem. This has come up once before and I contacted Spectrasonics about it and they asked the user to contact them about it. Unfortunately I’m not sure what became of it after that but I seem to remember that it was very particular to the machine in question - similar setup on another machine worked fine.
CPU or Memory would be the main two things I’d be thinking about. Have you my guide on tuning for audio performance?
Thanks for the tips so far. I have tried various settings for CPU support.
I’ve read through the audio guide - tried changing the paging size and location to an external SSD drive.
I still have crackling - predominantly when certain Omnisphere patches are used. Again, these same patches work fine in Cubase. I’ll try other patches for now.
Are you sure the exact same Omnisphere is being loaded into both hosts, and not for example a 32-bit on one and 64-bit on the other? Are you using the same audio interface with both hosts, and the same buffer sizes?
Omnisphere runs incredibly efficiently on my Cantabile setup, with similar hardware to you (except SSD) - one instance typically runs at about 10-15% load, with a 192 sample buffer.
I confirmed - running 64-bit Omnisphere in both Cubase and Cantabile. Same audio interface - Steinberg UR-22. Same buffer sizes. I think my slower hard-drive has something to do with it.
While you’re trying to sort this out I’d try disabling Cantabile’s multi-core support (Options -> Audio Engine -> Audio Threads -> 1) load just one instance of Omnisphere and play around with it’s settings.
Also, it’s been a while since I checked, but does Cubase have an extra layer of buffering - perhaps it’s running the same buffer size but multiple buffers? (just a guess)
I’m playing around with this 1 core setting. Maybe it helps a little bit. But still crackling whenever the page faults jump to 2000 to 4000 or so and the Load is around %130.
In Cubase are you playing Omnisphere in realtime or from a track? I notice you’ve got ASIO-Guard turned on which pre-processes audio tracks to prevent drop out - which you can’t do in realtime (ie: Cantabile can’t do this).
Also, do you have your PC power settings set to the High-Performance mode?
If you’re getting drop outs coinciding with Page Faults then I’m pretty sure this was the issue the other user was having - perhaps mention it to Spectrasonics and see what they say.
A note on asio guard
What it does is apply a buffer to tracks not monitoring live input.
Any track being played ‘live’ does not get any direct benefit from asio guard other than the relief resulting from all other, non ‘live’ components being buffered.
A side note: The downside of using a high setting for asio guard is that there is a dramatic slowdown in response when changing tracks as the system rebuffers everything.
High-performance mode makes a big difference. Here’s why:
In battery saving modes various parts of the CPU are shutdown and/or slowed down. You sit down at your keyboard and play that grand opening chord and suddenly the CPU has an awful lot of work to do. By the time the OS has re-awoken the CPUs the audio cycle is over and already glitched. Now imagine that all through your performance Windows is sleeping and waking your processors as load comes and goes.
For real-time audio processing, the CPU needs to be always ready - not in a few milliseconds time. This is why in my book I state that even if you ignore everything else in the book, don’t ignore switching to high-perf mode.
