In other words your average work load doesn't contain enough concurrent work items that are CPU bound.cgc said:My 2.66GHz MacPro doesn't use all four cores except on rare occassions (e.g. benchmarks, quicktime, handbrake, etc.) and even then it doesn't peg them all.
...some what a nonsensical statement...cgc said:What I'm most interested in is offloading OpenGL to a core, the GUI to another core, etc.
Threads of work are spread across available cores automatically. If a thread is ready to run and a core is idle then that thread will run on that core.
Aspects of the "UI" frameworks are multithread and will automatically utilize one or more cores (in some cases the frameworks increase the number of threads they use based on how many cores exist in the system). In other words the UI will already potentially use more then one core on a multi-core system.
The same can happen with OpenGL either now... say if the game developer for example utilizes one or more threads to calculate the game world state and a second thread to call into OpenGL to render that game world ...or by enabling the multithread OpenGL render (only available on Mac Pro systems at this time).
Of course that assumes that the tasks you run are CPU intensive enough to even begin to consume compute resources available to you in new systems... in the end you should measure overall throughput of the work load you want to do, not how utilized your individual core are when doing that work load.