Clip-based effects... is my workflow bad?

Started working yesterday on a "small" project.... one mic, multiple performers reading in front of an audience. So I am basically working on a mono audio file which is about 50 minutes long.
 
So, lots of different voices (each having their own tone and dynamics...)I decided to work on this file the following way:
 
1- Inserted an FX chain in that audio file (containing an EQ and a multi-band compressor tweaked to globally improve the audio...)
2- Began splitting that file into multiple clips ( one clip for each performer). Therefore, each of these new "performer" clips also has an instance of the above mentioned FX chain. This to then tweak the clip's FX chain to better suit the performer, based on the original FX chain's settings. (I have NOT bounced the clips to new audio files)
 
I now have about 15 clips with an instance of the original FX chain. Problem is, my CPU meters (in fact, core #1 mainly) is now spiking into the red and I am now hearing artifacts.
 
I thought that clip-based effects were demanding CPU resources only when the clip was "read". It seems not to be presently the case as I have problems no matter what clip I'm playing
 
So, in my case the FX chain seems file-based (and not CLIP based). (in SONAR X1 D prod)
 
Anybody else noted this behaviour? Looks like a bug to me, but my workflow might be wrong here...
 

I believe it's working as intended that splitting a clip with a clip FX (or chain) on it creates a new instance of the FX; it has to because clip FX are by definition independent to allow clip-specific settings. If you want/need to use one instance to conserve CPU, you'll need to put it on the track, and use automation to alter the settings from one clip to the next. Turning off Always Stream Audio Through FX in Preferences may reduce the load, but I don't think so; I think the FX will still be active, waiting for input.
 
Re-reading Michael's post, I see he's pretty much saying the same thing.
 
 

Thanks for the responses, guys.
 
Indeed, FX automation is a good way to go (and the one I'm using usually), but reading this "old" SOS column written by our friend Anderton made me want to try another workflow.
 
http://www.soundonsound.com/sos/oct07/articles/sonarworkshop_1007.htm
 
I also scanned X1's manual about clip FX. They're mentioned, but only briefly ... and without any comment about CPU ressources...

Seems like something for the bakers to fix/implement. I always assumed it worked that way too. Seems kinda dumb it wouldn't (just intuitively from the user's perspective, not hampered by any knowledge of the underlying technology). If Samplitude can do it, why not Sonar?

I would expect the clip's fx to automatically go on bypass once the clip is not being played. Normally this should release resources, right? Apparently something else is going on?

It's not necessarily easy to implement - for example, there have been problems in Samplitude when object-based effects induce a large latency, and there would be issues with reverb/delay extending past the edge of the clip.
 
I could have sworn, though, anyway, that there was some setting in Sonar to make clip based effects process only when necessary - are you sure that it's impossible?

I'm pretty sure delay and reverb fx already don't sound past the length of the clip (so you have to manually lengthen them), which is what has me confused. Or was that only during bounce-to-clip or something?