Your post is interesting to me. I have never used Sonar 4, but I do notice that Sonar 5 seems to use more CPU than Sonar 2 for the same project. See the following thread:
https://homerecording.com/bbs/showthread.php?t=175656&highlight=cpu+sonar
There seems to be some debate about this because the CPU meter may not be accurate. However, it does seem accurate to me in one important respect: as it gets close to 100% I get dropouts!
One thing I have noticed is that with Sonar 2, the CPU meter was basically at zero until I started playing the project. However, in Sonar 5, a complicated project might have the CPU running at 30-40 per cent before I even hit play. The plug-ins seem to use CPU even when the audio engine is "idling"
Other people report that they find Sonar 5 to be as efficient or more so than previous versions. I never had any dropouts before Sonar 5, but on the other hand my projects are getting more complex with time--so more plug-ins, synths, etc.
One thing I do notice is this. I used to record at 24/96. When importing some of these older projects into Sonar 5 to re-mix, add plug-ins, etc., I find my system doesn't handle them well. With 16/44, I'm not having any practical problems.
You could always just continue to use Sonar 4--you would have access to most of the Sonar 5 plug-ins, with the exception of V-Vocal. Or, you could be liberal with your use of "Freeze".
I am still intrigued by this efficiency issue. I think I will post a poll.