I have two machines with two processors each. Presently running boinc-4.13. One machine has hyperthreaded XEON processors and the other has Pentium IIIs.

It sometimes happens that even though there are plenty of available work units (I am running setiathome and climate-prediction), one (or more) of the processors goes idle for a noticeable time. If I kill the boinc client, that usually, but not always, kills the application process(es). If I then restart the boinc client, an application process starts for each processor.

Is it normal that sometimes the processors are not used even though there are work units available? Should I have to keep an eye out for underworked machines and restart boinc on them? Is there anything I should look for?

One machine runs Red Hat Enterprise Linux 3 with kernel-smp-2.4.21-20.EL and the other runs Fedora Core 2 with kernel-smp-2.6.8-1.521. Both are running /boinc/boinc_4.13_i686-pc-linux-gnu.

---