Benchmark (was: Re: Junior kernel hacking)

Jean-Marc Zucconi jmz at dalai-zebu.org
Wed Jan 14 18:33:37 PST 2004


>>>>> Matthew Dillon writes:

 > :I have run bytebench on a machine running FreeBSD 5.2 and
 > :DragonFly (this is old hardware: dual PII @333MHz). I am going to
 > :install FreeBSD 4.9 on a third slice to have another comparison
 > :point. Here are the (surprizing) results:

 >     Hmm.  Well, while the numbers are impressive 5.x can't be
 >     doing as bad as some of them indicate.  There must be some
 >     basic differences that are causing some of these results.

 >     Be sure to compile 5.x without DIAGNOSTIC, INVARIANTS, 
 >     or WITNESS.  Be sure that the kernel is being compiled -O.
 >     Doubly be sure that WITNESS support is not on.  INVARIANTS 
 >     and WITNESS *are* turned on in 5.x's GENERIC.

OK, I will check this as soon as the benchmark on 4.9 finishes. I
think I have INVARIANTS but not DIAGNOSTIC nor WITNESS. kernel and
world is always compiled with -O (and no extra flags). I will also
test on 5.1-R as I have it on the 4th slice. I will post the results
tomorrow.

Jean-Marc

-- 
Jean-Marc Zucconi -- PGP Key: finger jmz at xxxxxxxxxxx [KeyID: 400B38E9]





More information about the Kernel mailing list