My system must be blessed as my cores are almost always equally loaded.  I don't have any projects of any size of my own, so I just loaded the Sonar factory projects (which aren't huge either).    With Cori Yarchin's Floating, my 8 virtual cores are almost exactly even and rise and fall together.  Core 1 may be a hair above the others, but it's also sometimes lower.  They all hover around maybe 10%, rising at bit on certain passages to no more than 20%.
 
What I find interesting is that the load on the cores is as much, and even higher, with playback paused (still constant across cores) than in play where the load rises and falls.
 
[EDIT :  with the 4 synths unfrozen, core 1 rises a bit above the others, maybe 1/4 more than the others.]
 
Playing "Javier Colon" (all unfrozen), core one is also a bit higher than the others.  2-8 are evenly loaded.
 
I have wifi on and Norton in the background so I question the idea that Windows itself (win 7 at least) places a greater load on core 1.
 
I looks like soft synths are what puts demands on core 1.
 

Paul P
 
My system must be blessed as my cores are almost always equally loaded.  I don't have any projects of any size of my own, so I just loaded the Sonar factory projects (which aren't huge either).    With Cori Yarchin's Floating, my 8 virtual cores are almost exactly even and rise and fall together.  Core 1 may be a hair above the others, but it's also sometimes lower.  They all hover around maybe 10% rising at bit on certain passages to no more than 20%.
 
What I find interesting is that the load on the cores is as much, and even higher, with playback paused (still constant across cores) than in play where the load rises and falls.
 
Playing "Javier Colon", core one is a bit higher than the others for some reason.  2-8 are evenly loaded.
 
I have wifi on and Norton in the background so I question the idea that Windows itself (win 7 at least) places a greater load on core 1.
 

Hi Paul, Do you use AD2?

PassiveDrift
Do you use AD2?

The Javier Colon demo uses AD1 and the load wasn't extreme.  But :  I just loaded one of my (daughter's) songs to which I added AD2 and now core 1 rises to around 50%, way above the others (I'd never noticed).  Freezing it brings core 1 down, but it still remains a bit above the others.  I guess the Cori Yarchin demo is so even partly because the drums are all audio tracks.
 
So AD2 could benefit from sharing its load across cores.
 

Paul P

PassiveDrift
Do you use AD2?

The Javier Colon demo uses AD1 and the load wasn't extreme.  But :  I just loaded one of my (daughter's) songs to which I added AD2 and now core 1 rises to around 50%, way above the others (I'd never noticed).  Freezing it brings core 1 down, but it still remains a bit above the others.  I guess the Cori Yarchin demo is so even partly because the drums are all audio tracks.
 
So AD2 could benefit from sharing its load across cores.
 

yeah AD2 is my main problem, but ok now with this core spread. Maybe XLN could look into that.
 
Thanks for the verification :)

PassiveDrift

Jyri T.
I, too, have this problem. The worst culprits seem to be the Slate Digital virtual emulation plugins and some DMG Audio ones.
I have tried all the tricks in the book (including those mentioned in this thread) but no solution yet.

Are you talking about the The Slate Virtual Console Collection? I tested that today on 12 tracks plus on the bus, It worked fine BUT I have the 64 bit upgrade. The upgrade was free but I had to ask Slate for it. I don't know if it's still a free upgrade.

Never mind. This was an IUIQ (insufficient user IQ) error. I had accidentally limited my CPU to max. 5% in the power management profile settings.
Although the first core (actually thread) DOES hit higher than others, it's still perfectly usable after changing the power setting...
Sigh... I'm a good candidate for the Sonar Forum Idiot Of The Month prize.

PS. The first CPU spiking seems to be less of a problem with the two latest Sonar updates, especially with Manchester. At least with brand new projects.

I set my "MixThreadCount" to 4 and set the Enable set threaded ideal processor to "False". Now the activity in the CPU Meter is spread out pretty evenly across all the cores (eight showing for me). I had been dealing with the spiking first core for months. I'm so happy I found this fix. Now, can the bakers explain why it was happening in the first place? And why it never used to happen but now does to some us?

I am curious to try this experiment now, and also wondering if this OS specific.

It might be that there's some error in the logic that estimates how to distribute the load, and that Sonar wrongly puts something on the busy core when it would have been better on a different core.
 
Or, it could even be the other way - that Sonar is avoiding other cores because some background processes are using them, making them less suitable for real-time audio.

It would also be interesting to see if, when you guys have solved a problem by limiting the thread count, whether that (a) gives you a benefit across all your projects, including those with different plugins, and (b) whether that still gives you a benefit at higher loads - eg. 2, 3, or more instances of AD2.
 

Thanks everybody for the input, this was very useful, I have been struggling with core "one" or "zero" maxing out even on plugin modest projects with my 12 core super duper i7. So I played with "MixThreadCount" and Enable Threaded Ideal processor set to "False" and core usage is much more even now. But I feel uncomfortable having no clue what I was doing.  Can anybody explain what are those parameters?