I did write a thorough post with lots of detail but it timed out my login and I lost the entire post!!! Ahhhgg. So here is a simpler summary.
I use VEPro7 all the time. Big orchestral projects and a large template.
I upgraded to C10 from C9.5 last week - but I have to go back to C9.5 - here’s why:
With Cubase 10, as soon as I record-arm a midi channel to send midi to an instance of Vienna Ensemble Pro (VEPro7 v 7.0.851) the CPU meter in VEPro goes up considerably, even with no notes being pressed or midi CC data being sent.
Arm 2 midi channels to the same instance and it goes up some more. Even dormant it is using more CPU than in C9.5
The result is that with a section of 1st violins (9 VI’s being sent data from one channel of Cubase) the audio breaks up and the VEPro meter hits 100% in no time.
In Cubase 9.5 this just does not happen. I can arm as many record tracks [to the same instance] as I like and there is absolutely no discernable change in the CPU figures noted on the VEPro7 tabs.
And I can run an entire string section (27 virtual instruments in my case) simultaneously and polyphonically and there is still plenty of headroom. No audio breakup whatsoever.
All this is with ASIO guard turned off for VEPro - in both sets of tests - [turning ASIO guard on (naturally) just makes everything much worse.]
So Cubase 10 seems to be sending spurious unnecessary data of some kind to VEPro which is just killing the CPU handling.
This is new behaviour because Cubase 9.5 isn’t doing it.
All the tests were with the exact same projects on the same machine, same version of VEPro and so on. All the settings are also exactly the same.
It does not matter whether the test project was originally created in C9.5 or C10 - the results are exactly the same.
Hardware: Mac Pro late 2013 model - 12 core
VEPro7 v 7.0.851
Cubase 9.5.40 build 259 (works fine)
Cubase 10.0.30 (breaks VEPro instance CPU barrier very quickly)
Mac OS 10.13.5