.. also, if you take a look at the ktr output, the CPU timers between CPU 0 and CPU 1 are slightly different. schedgraph complains quite loudly. :-) Is that acceptable/possible? Adrian