I've noticed some odd results in a home-grown statistics program running
on different AIX 5.3 TL 5 SP 4 boxes. In a group of four LPARs, three
were built with ML 3 and upgraded to TL 4 SP 4. The fourth was built
straight to TL 5 SP 4 based on an upgraded version of the same SPOT I used
to build the other three...

On the fourth LPAR, it looks like I'm seeing 2.5 times one of the metrics
I should be. For example, my 4 processor box with an LPAR that can't go
beyond 3 physical CPUs is maxing out at 7.5 CPUs...

The same box seems to work fine with both topas and nmon, so perhaps
they're doing the calculations in a different way...

Thanks in advance!

-Chris