Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!samsung!zaphod.mps.ohio-state.edu!sol.ctr.columbia.edu!ira.uka.de!smurf!flatlin!bad
From: bad@flatlin.ka.sub.org (Christoph Badura)
Newsgroups: comp.unix.sysv386
Subject: Re: Take out memory to speed up kernel build!?
Message-ID: <1253@flatlin.ka.sub.org>
Date: 3 Oct 90 01:48:49 GMT
References: <34417@cup.portal.com> <4133@altos86.Altos.COM>
Organization: Guru Systems, Karlsruhe, Germany
Lines: 23

steve@altos86.Altos.COM (Steve Scherf) writes:

>In article <34417@cup.portal.com> ts@cup.portal.com (Tim W Smith) noticed:
>>[that kernel build time dropped from 7.5 minutes to 5 when he took a
>>memory card out of his system.]

>Can you be more specific about the environment under which you tried this?

When I brought up Unix for the first time on my home System (16 MHz
SX) I used a spare 2MB extended memory card in the System. That gave
me 4 MB but the System was *very* slow. I then removed the extended
memory card, leaving me with only 2MB, and the box screamed (well,
sort of). With 2MB RAM the system was 3-5 times faster.

The explanation is simple. Xenix noticed the additional RAM and
used more of the mainboard memory for the buffer cache and put
virtually all processes on the RAM card. But the card was running on
an 8MHz IO-bus with atleast 1 Waitstate whereas the block buffers
where using the zero-waitstate RAM on the Mainboard at 16 MHz.

-- 
SVR4 is an adventure; if you win		 |	Christoph Badura 
 you find you're playing VMS. -- Richard O'Keefe |	bad@flatlin.ka.sub.org