> Hi Jaromir, > Hi Jim. > I just ran a little test on my 4-core laptop. > > Using top with some fprintf statements, taskset for cpu affinity, > cpufreq-set to force a given frequency and cpufreq-info for > verification, here's what I found. > > The kernel tics recorded for /proc/stat line one AND each individual > cpu line are virtually identical when running at either 800MHz or > 2.10GHz. > > Does this make your problem go away? Actually I did my own tests on my workstation with: Intel(R) Xeon(R) CPU W3520 @ 2.67GHz The test application was producing a fixed amount of load triggered by the system clock each second and the results were surprisingly opposite: 1.) with frequency set to 2.66GHz the load was approximately 34% 2.) with frequency set to 1.73GHz the load was approximately 52% Whilst the load should be 34% in both cases ... I tested that with procps 3.2.8 and procps-ng 3.3.2 and both of them returned the same result. > > Jim > Jaromir.