I'm up to 69 tracks on "Arctic Emergency", my highest track count to date.  Saturday I purchased "Juggernaught" from ISW and in an all-nighter, added 25 tracks of brief but potent cinematic "whoosh bangs" as Daniel James fondly calls them... as it was morning and I had Sunday plans, I quickly added 25 instances of CA-2A leveler to levelize the whooshy's and dumped a mix to wave/mp3's while I still had a shred of lucidity within me...
 
Tonight I returned to the project and noticed high cpu levels (50 percentish) on all 12 threads and spiking on no. 1.   Actually, I noticed it after multiple dropouts stalled plays! Playing back at 10ms buffers, longest allowed by Forte.
 
CPU spiking hasn't happened since the upgrade from i7 960 to the 4930.  Copied the song, and started deleting things added since the last mix.  As soon as I deleted all 25 new CA-2A's, CPU levels returned to normal...about 30% on the first thread, 10-20 percent on the rest.
 
So I guess the pro channel is to be used only when needed. Even with my beefy system, can't just lay it on everything, even if it's just the leveler... btw, just sitting idle the levelers really ramped up my thread loads...
 
Is there any data on cpu hit of the various moduals?
 
Is something like FF Peq more efficient then the prochannel flyout eq?
 
Am I doing something wrong or just foolishly unaware of reasonable pro-channel constraints now that I'm getting into high track counts? I guess I can still apply it to the sub mix busses... after roughing out levels on the whoosh bangs... but can't indiscriminately use modules
 
did go back and manually tweak levels, as a good little surgeon...and don't need them 25 levelers now.  One thing I just realized in cinematic sound work is that consolidating the itty bitty sound clips (waves) for fewer tracks save lots of cpu overhead.
 
Of course you can tell me I've got it all wrong...just stumbling around in a sea of ecstasy...
 
 

10ms is the highest the Forte will allow? When mixing with that many effects/tracks higher latency will relieve a lot of the strain on your CPU, but putting effects on buses rather than tracks can reduce the CPU load a lot as well. Another thing to consider is bouncing any intensive synths to new audio tracks and archiving the originals (or freezing those tracks). The CPU is made a lot happier when it is playing audio rather than generating it each time.
 
Even with a beefy system, it is still prudent to be conservative. Not everything plays well in the sandbox with each other.

Just trying to understand the limits.... Obviously, can't apply lever's to every track! Unless someone comes on to say, sure you can and your system is all fked up!  (no, don't say that! I wub my new system!
 
But being wiped out, sleepy... and just applying 25 levelers taught me not to be lazy!  
 
Would luv a relative cpu hit ranking of the different modules but, why I don't know...

70 tracks and 25 instances of prochannel and you wonder if your system is up to snuff?  That seems a lot to me - and apparently your system, too.  Why not drop the latency of the forte, too, and really give your cpu a work out?  But kidding aside, 10 ms seems high, which indicates a problem w/ the drivers, not your specs.  If you are really intent on putting a large, unnecessary load on your computer, you need look at the audio interface first.  but maybe the Forté is problematic - I'd look at the Focusrite forums and see.
 
But you most likely don't need a comp on each wooshbang.  As suggested above, a wooshbang buss, with the SSL buss comp would (probably) sound as good.  I'm sweating just thinking about going through and adjusting 25 comps on what is a background sound.  That is 30 minutes or so of setting individual comps, even after you find the "correct" setting for the first one. I guess you could save the preset as Wooshbang, but the only reason to use a separate comp is to adjust each one individually.  I see a whole afternoon gone there. 
 
But, more power to you.
 
@

I think it's fairly easy on the cpu. It's not just a leveler, it's an emulation of a vintage compressor, even if they called it a "leveling amplifier" at the time. I would be surprised if you would get better performance inserting 25 instances of another recently launched compressor.
A fast CPU doesn't mean you shouldn't consider proper routing and FX use.

These effects are not CPU hogs. You just need to raise your ASIO buffer or WDM slider. The lower your ASIO buffer or WDM slider, the more your CPU works. The higher your ASIO buffer or WDM slider, the less your CPU works.  
 
So with this information,  you can conclude that bigger projects may need a higher sound card driver setting. Thats all it is.
 
FYI: I can have 100 plus tracks, 70 instances of pro channel and 50 vst effects and none of my cores will spike. Why, because i ahve my sound card driver setting correctly for that Sonar project

Sorry guys...I've been away.  I only posted to get insights into cpu load with pro channel.  I would never end up with that configuration, was just headed off to bed after a grueling 24 hours and just slapped um on, half delerious/lack of sleep, having a blast with the new Juggernaught whoosh bang wave files...which I went crazy with, as one tends to do with a new girlfriend! I posted out of noticing the impact on cpu...of using pro channel's CA-2A on lots of tracks.  
  
So, I realize is was not something I would want to do in a real world situation but was surprised at spiking, as I understood that pro channel was low impact.  I do understand/am working hard at honing good mixing skills rather than using crutches to mask doing it right.  Especially as I'm getting more into orchestral/impact wave files to round out a song.
 
So, what I'm left with is:
1. High track counts strain cpu... even if little is on them (a cello bit for the bridge, for example). If so, this kinda sucks as orchestral sound design seems to require lots of unique intruments which come in and out sparsly...
 
2. In doing "whoosh bangs" I should consolidate those waves on as few tracks as possible to reduce cpu load, and automate volumes as needed of each chunk/wave
 
3. Use many submixes for the above and process those if needed
 
4. Route any verbs/delays/whatever to their own subs, using individual sends of course...but my choice is either send from each track or from submix bus (if just an overal verb ambience is needed)
 
Just trying to sort out some best practices for large track counts; am a bit concerned that CJay can make use of large pro channel/effects on large tracks " Why, because i ahve my sound card driver setting correctly for that Sonar project" 
 
I do believe I am also doing so...adjust buffer size according to what I wish to do and always switch to highest buffer (10ms) when playback/mixing... forte uses different nomenclature than others, for some reason so that may be the confusion.  At 10ms buffer on forte (960 samples, it's highest setting), my latency is 51.2 msec, running 96/24.  I generally record at 1ms buffer at a much lower latency of 6.2 msec. when latency bugs me...
 
I will check for driver updates as the forte is relatively new to focusrite... My other intention was to see if I had a glitch in my system, because high use of prochannel shouldn't cause spikes... I still don't know if that may be the case.
 
 
 
 

mettelus
10ms is the highest the Forte will allow? When mixing with that many effects/tracks higher latency will relieve a lot of the strain on your CPU, but putting effects on buses rather than tracks can reduce the CPU load a lot as well. Another thing to consider is bouncing any intensive synths to new audio tracks and archiving the originals (or freezing those tracks). The CPU is made a lot happier when it is playing audio rather than generating it each time.
 
 

Forte uses ms instead of "samples" so, 1 ms: 96 samples.  Don't know why they did that.
 
All Whooshbangs were "waves", not synths... the nice thing about Juggernaught...500 wave files included so it can be used direct as well as within' Kontakt for more elaborate processing...I just dragged a slew of um in to various key points and played around.  It was my first date with the new hottie! I went a bit crazy (and had fun)

10ms=96 samples can't be right. You mean 1ms=96 samples, then it makes sense (96KHz).
Also, it doesn't really matter if very little is on the track because as far as I understand, Sonar is always processing the VSTs even if no audio is running through them. If it's a set and forget deal for you then you can freeze the tracks. Surely, freezing 25 PC tracks with 25 CA2As will lighten the load on your system.
 
I think another thing putting strain on your system is that you're doing all this at 96KHz. I know some people really feel some need for this but 44.1/24 or even 48/24 would be easier for any system. The advantages for using a higher samplerate are mainly in tracking quality. This sounds like a mostly in the box production. Meaning most of your tracks are already artificially upsampled. Just seems like a waste of CPU to me.
 

Actually high track counts don't stress a CPU as much as they challenge disc through put. what will stress a cpu with track count is if they're at a high sample rate like 96khz.
 
just because you have a beefy computer doesn't mean you can forget about the judicious use of plugins and having you drivers setup properly, with correct latency for the  job. If you're mixing raise your latency.
 
Re-read cj's post.