On Fri, 26 Sep 2003, Michael K. Johnson wrote:
One possibility, especially if you were comparing to an up2date
kernel
from test1 (only a week's worth of changes, basically), would be that
the scheduler interactivity changes that were just accepted into Linus's
tree in the 2.6.0-test kernels were backported. (A think-o in that
backport is what caused the SMP instability and has now been fixed in
the packages at
http://people.redhat.com/johnsonm/severn/)
We at least think we can really notice the difference.
At least on this desktop, it seems faster, but "jerky" -- operations
normally go smoothly and quickly, but sometimes they pause mid-operation for
a second or so (Athlon 1700+, 768 megs, just running ~12 xterms, mozilla,
and xmms)
later,
chris