on some of my faster machines, the load average is often quite wrong when there is a single process using a full cpu. for example, the load average on this machine is 1.000 but sysstat claims it is 0.002.
chula; 8.burncycles & sleep 60
chula; cat /dev/sysstat
0 16041 1216771 646 305 0
0 2 99 0
1 30658 1073202 13566 335 0
0 0 99 0
2 19686 1069572 2433 1294 0
0 0 99 0
3 19441 1070607 1160 311 0
0 0 99 0
4 22342 1071838 2210 1068 0
0 0 99 0
5 14647 1133929 28 22 0
0 0 99 0
6 18757 1070358 1576 208 0
0 0 99 0
7 20882 1071157 6222 5636 0
0 0 4 0
starting a second cpu-hogging process will make the load average 2.000.
- erik
