I had peaks after disabling MP on my LARGE projects but that’s because you obviously lose horsepower when you turn off the MP. That is different from the freezes. Are you testing with a heavy project?
My workloads are reasonably heavy - but note that I do NOT use any virtual instruments and rarely do studio style live tracking or overdubs.
I sometimes use Midi in the tests to run a simple control surface via USB.
I have a dummy Midi track defined in all projects to allow me to export marker information to my own locally written Cue file generation program, but the track never has any midi IO or other events in it and is never edited.
My current reference EDIT is a 12 track mixdown/mastering combined project of a 70 minute live ethnic music concert where I clean up things and break the performance up into patter and music sections and edit it to a Stereo .Wav master and .Cue file for release.
To do this I use 1-3 plugins for EQ and compression/limiting on nearly every track.
I use two EQ plugins, MBC, Limiter and Dither on the Stereo output to master the mix.
I do not use third party or any of the new channel strip plugins other than EQ to keep the project exactly compatible with Cubase 6.5 so that I can move back and forth between them for comparison.
Currently, editing consists of working through the entire mix trimming levels, tweaking cue points and fiddling with the EQ and panning.
I have not added any automation to any of the test projects yet.
I’m working with an Edirol FA101 8 channel firewire interface 44khz/24 bit for the mixdown tests (A great reference device since it uses Class IO and has no drivers on the Macintosh).
When testing with control room active: I have 3 stereo outputs defined - one for main mix, one for control room and a second control room output with some plugins to correct one of my reference headphones.
Without Control Room: I use one stereo output pair for the Main mix / control room.
My Reference Recording tests use a Phonic HB 1024U - which also uses the Class drivers and so needs no custom drivers. For these tests, I’m using the Firewire interface (not the USB 2.0) at 44.1khz/24 bit to record an identical test audio stream from a radio station to 18 tracks for 1-2 hours per test, using the 2 channel return as the stereo control room/ main mix output.
I’ve also done some testing with the Edirol and Phonic as an aggregate device with 24 channels of input, and a control room and two auxiliary headphone outputs split across the two devices.
In addition to monitoring for Processor overload alets, VST Performance and Apple’s Activity Monitor, I check for dropped samples across all tracks after each recording test.
Like you, I’m seeing the usual need to increase buffer sizes to accomodate heavier VST processing loads in the live recording tests - especially as I add features like the control room, meter bridge and open plugin windows - but things are scaling just as they always have here in Cubase 4, 5 and 6.5.
I get stable results (no monitoring breakup or processor overload warnings) with buffer size 256 for the recording tests, increasing that to 1024 for the aggregate device test - this seems to always be needed if I aggregate output devices - otherwise lower latencies still work. For the Edit tests I use 512 with no playback breakup. (of course when I’m actually recording a concert, I’m usually more conservative in these settings).
The “Freeze” scenario is something different and I am only seeing it with Cubase 7 when MP is turned on, and then it seems little related to workload - although it seems to get somewhat worse as workload increases. I’m seeing it even on projects with a single stereo track and no plugins.
hope this info isn’t too boring…
M