On 12/10/2012 10:31 AM, Jon Chiappetta wrote:
I realize that these numbers seem very minor but I did notice a much
faster drawing performance when
firstboot was started compared to the v5 version of Fedora. Also, it
takes away any excuses or doubt
that we're slower than raspian because we're not v6hl. In addition, if
v7hl gets moved over to primary
one day, then Seneca can still run a farm that has v5 and v6 releases of
Fedora. Thanks for your time
and testing, I appreciate it as always.
It looks like only bzip2 ran for long enough to make a valid comparison.
I'd be interested in reproducing any of the benchmarks used here:
The above shows the difference between armv4 soft float vs armv6 hard
float. Presumably armv5 soft float to armv6 hard float will end up
somewhere in-between these two points. The big question is where will
it end up? If armv6 hard float is substantially similar to armv5 soft
float that tells us that the raspbian benchmarks primarily benefited
from the move off of armv4. If on the other hand we get significant
benefit from armv6 hard float, we know either the hard float or the
v5->v6 move was very important. Appreciate all the hard work you've put
into bringing armv6hl online. We're just on the cusp of knowing how
important it is to continue its pursuit. Keep going!
Brendan Conoboy / Red Hat, Inc. / blc(a)redhat.com