Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 5 Nov 1999 11:03:17 -0800 (PST)
From:      "Rodney W. Grimes" <freebsd@gndrsh.dnsmgr.net>
To:        ejs@bfd.com (Eric J. Schwertfeger)
Cc:        adams@digitalspark.net (Adam Strohl), doconnor@gsoft.com.au (Daniel O'Connor), darrylo@sr.hp.com (Darryl Okahata), freebsd-smp@FreeBSD.ORG
Subject:   Re: Dual Celeron + FreeBSD?
Message-ID:  <199911051903.LAA57990@gndrsh.dnsmgr.net>
In-Reply-To: <Pine.BSF.4.10.9911051041430.16049-100000@harlie.bfd.com> from "Eric J. Schwertfeger" at "Nov 5, 1999 10:47:30 am"

next in thread | previous in thread | raw e-mail | index | archive | help
> On Fri, 5 Nov 1999, Rodney W. Grimes wrote:
> 
> > > I've done make buildworld's on both my dual Celeron and my dual PPro (the 
> > > ones with 512K cache). The difference between single and dual celeron is
> > > minimal, about 10%.  On the dual PPro machine, the speed improvement,
> > > using the same disk subsystem, was 80%.  Yes, on processes that aren't
> > > memory intensive, dual Celerons rock.  In fact, on most things, I see
> > > closer to 40-50% improvement with dual Celerons, the make buildworld is
> > > rather memory intensive.
> > 
> > Actually make buildworld is disk intensive... SMP plain out does not
> > seem to help it much, unless of course you run a non-standard make
> > world with -pipe, which then does make the memory bandwidth demand
> > higher, and if both sides of the pipe just happen to get split accross
> > 2 processors it causes the small cache to be ineffective and the memory
> > system to be a major stall point.
> 
> Can't remember if I was using -pipe, this was a default 3.2 install,
> except that I'm pretty sure I was using soft updates. I may have been,
> because the HD lights would only flicker every few seconds, so it
> definitely wasn't disk bound. I had /usr/src, /usr/obj, and /tmp all on
> seperate spindles (each on a 4GB 7200 RPM SCSI drive).  Given other
> people's comments, I'd really expect more than the 10% that I saw.

Soft updates shifts the bottleneck from disk to memory.  We run our
baseline comparison testing using unmodified in anyway installs from
CDROM of the latest version of -release.


-- 
Rod Grimes - KD7CAX @ CN85sl - (RWG25)               rgrimes@gndrsh.dnsmgr.net


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-smp" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199911051903.LAA57990>