Date: Wed, 22 Mar 2023 10:10:43 -0700 (PDT) From: "Rodney W. Grimes" <freebsd-rwg@gndrsh.dnsmgr.net> To: sgk@troutmask.apl.washington.edu Cc: Mark Millard <marklmi@yahoo.com>, George Mitchell <george+freebsd@m5p.com>, FreeBSD Hackers <freebsd-hackers@FreeBSD.org> Subject: Re: Periodic rant about SCHED_ULE Message-ID: <202303221710.32MHAhe9047582@gndrsh.dnsmgr.net> In-Reply-To: <ZBsxUYQwqHvuIdpe@troutmask.apl.washington.edu>
next in thread | previous in thread | raw e-mail | index | archive | help
> On Wed, Mar 22, 2023 at 09:20:27AM -0700, Mark Millard wrote: > > > > Giving folks a way to know they are repeating your tests > > appropriately, could give interested folks a way to answer > > their own questions. > > This has been an issue for years (and now stretching into > decades). It is trivial to show the problem with any > numerically intensive MPI program. I've done this a few > times, and reported the issues. Search the mailing list > archives, e.g., > > https://lists.freebsd.org/pipermail/freebsd-hackers/2008-October/026375.html > > It appears to be (or was) an issue with cpu affinity. > > Caveat: I haven't tested this in a long time. I simple use 4BSD. I dont even try ULE any more. I just used 4BSD, as did bde@freebsd.org, ULE seems to suck when your have interactive use and compute bound on the same box. I have seen interactive in the past take seconds to echo a command. IIRC ULE and zfs in a memory contrained environment dont play nicely togeather either. +1 on the return to 4BSD as the default scheduler -- Rod Grimes rgrimes@freebsd.org
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?202303221710.32MHAhe9047582>