No. They are total differenr concepts. Let me explain more -
1. Elapsed time = end time - start time (profiling)
2. CPU time of function : captured samples of function * interval . For example, if you got 100 samples in f1(), and interval is 10ms as default. So CPU time of f1() = 100 * 10ms = 1s
3. CPU Utilization : time period from function's entry & exit, it lasts 2s. As I said f1() took 1s in one core, so CPU utilization is 50% (probably instruction was suspended, IO waits, etc)
4. Same situation for multicore, if the system is 2 cores, Full utilization is 200%, follow the point 3, meanwhile if
a) Other function ran 2s on another core, so CPU utilization is 3s / 2s = 150%
b) Other function ran 1s on another core, so CPU utilization is 2s / 2s = 100%
c) No other function run on another core, so CPU utilization is 1s / 2s = 50%.
Please observe timeline report to view these data.
Determine CPU utilization(%) from CPU time given by VTune
Currently i am working with Vtune. I am trying to calculate CPU% utilization give the "CPU seconds used" and the elapsed time.
Is this write : CPU% = cpu time(secs) / elapsed time. ?
e.g. in 2 core machine
cpu sec =5.6 , elapsed time = 8.4
cpu% = 5.6 / 8.4 .
Also how to map cpu time and utilization in multicore environment.
Thanks,
rupinder