I dont really know how to pull out the cpu utilization in a windows machine, but then you wouldnt be using one for grid computing would you???!! a UNIX/linux machine would probably be easier since the 'top' command gives the overall useage and gives a breakdown of the useage for each process running. This could be determined by a shell script that manipulates top, or just pull apart the source code for 'top' to get what you want.
I'm guessing you need the cpu utilization to determine which machine to ship the next batch of processing to. In which case do you need a graphical interface to the cpu utilization? probably not - just code it in C or if you need to C++, forget any visual aspect to it unless explicitly required.
Just some other thoughts:
> When allocating processing make sure you use an average cpu utilization measured over a period of time rather than the cpu utilization at an instant since an intantaneous check might see one CPU at 100% and another at 20%, this would result in the processing being shipped to the 20% CPU whose process might run over a longer period than the 100% CPU who processes very short jobs at sparse intervals and could actually be the best option.
> You might also want to take note of network bandwidth. Even if a CPU not being used too efficiently if it has very little network bandwidth available then it may be quicker to ship the job elsewhere.
hopefully some of that is useful even if it doesnt really anwer your question!
code for cpu utilization 
Linear Mode
