low(er) disk performance with sched_4bsd then with sched_ule

Joseph Koshy joseph.koshy at gmail.com
Sat Sep 17 19:05:19 PDT 2005


> I don't know the disk codepath very well, but the samples look 
> little suspect.

They do look odd.  For one, there are entries for the p4_*_pmc
functions which would never get called on a P-III box.

Oliver, after you sync'ed up with RELENG_6, did you keep the
sources for pmcstat(8) sync'ed with current?  If not, that
would explain the skewed profile.  Yesterday I MFC'ed a
few bug fixes so could you run the last step (pmcstat -g -R 
/path-to-sample-file) with today's pmcstat(8)?   The file
"src/usr.sbin/pmcstat_log.c" should be at rev 1.3.2.1.

-- 
FreeBSD Volunteer,     http://people.freebsd.org/~jkoshy


More information about the freebsd-current mailing list