Federico Giannici wrote: > > > I have just switched to a multiprocessing kernel (3.9-stable i386) with > a dual core Athlon 64. > > I noticed that "top" command now have two "CPUx" rows, one for each CPU. > But "iostat" has only one "cpu" column. > > Question 1: are the iostat's cpu values a mean of the values of both CPUs? > > Regarding the "load average", obviously, even with two CPUs, there is > only one value. The problem is that it seems that it's values are almost > the same of the uniprocessor kernel. > > Question 2: given that with with double CPUs there is half the > probability to find a busy CPU, shouldn't the load averages be about the > half of the previous values?
Only if waiting for a CPU were the only cause of waiting. It should be possible to contrive something where the load averages are double with two cpus. Load average is easy to compute, but cannot really mean much.
