ken mays wrote:
They reported the apps they used were 'out-of-the-box' with no tweaking
on their (i.e. Phoronix) part. Whatever was on the CD is what they used.
A few of the 'optimized libraries and apps' are over at Blastwave.org
so we can easily retest the benchmarks to see if GCC optimizations were the
case.
~ Ken Mays
It doesn't really address the question of whether there was bias in the
selection of tests. As I mentioned earlier, the tests that were added
since their last testing were the ones which OpenSolaris did the worst
on. Four of those tests were against GraphicsMagick which only used
OpenMP if the binary was compiled under a newer version of the gcc
compiler than that which ships with OpenSolaris. At first blush, the
other packages selected have similar issues.
I agree with the concept of testing similar applications under the two
operating systems as being a fair way to compare them, IF, the
applications selected are a fair representation of what an average user
would experience rather than being cherry picked to give the results the
tester wanted. I'm not sure the tests used are a fair representation of
the average user experience though.
Regards,
Greg
_______________________________________________
opensolaris-discuss mailing list
opensolaris-discuss@opensolaris.org