Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!ames!think!samsung!cs.utexas.edu!utastro!hgcjr
From: hgcjr@utastro.UUCP (Harold G. Corwin Jr.)
Newsgroups: comp.sys.ibm.pc
Subject: Re: Free Utilities
Summary: Refresh rate: experiment!
Keywords: RAM refresh rate
Message-ID: <4930@utastro.UUCP>
Date: 16 Mar 90 19:40:17 GMT
References: <72@demott.COM> <313@pallas.athenanet.com>
Organization: U. Texas, Astronomy, Austin, TX
Lines: 34

In article <313@pallas.athenanet.com>, dgl292@pallas.athenanet.com (Doug Lee) writes:
: In article <72@demott.COM> kdq@demott.COM (Kevin D. Quitt) writes:
: >    REFTEST allows you to adjust the DRAM refresh rate on your machine
: >to optimize performance.  Based on its numbers, Norton's si, and my own
: >benchmark suite, my machine now exhibits a 4.1% speed increase.  BTW,
: 
: CAREFUL!!!  Adjusting the RAM refresh rate can, indeed, produce a speed
: increase.  However, changing it too much can, in my experience, cause some
: WEIRD problems.  On my machine, which is a Zeos 12 MHz 286, a refresh rate
: which was too slow caused a slow death of memory (i.e., bits started flipping
: at random!!!).  The trouble is NOT something you are guaranteed to see
: immediately; sometimes I could run for several minutes before noticing
: strange behavior.
: 
: I have no way of knowing how other machines will respond to an overly slow
: refresh rate, but I would advise caution to anyone who tries to slow it

The caution above may be true, but simply experimenting with longer and 
longer refresh rates will tell you what your machine can handle.  I have an 
8-MHz 286.  With the default refresh rate of something like 15 millisec, I 
was getting around 1600 dhrystones.  I found that I could slow down the 
refresh rate to something like 0.1 sec and get 1800 dhrystones.  But if I
went beyond that, the machine did indeed do odd things.  Still, I've been
running at the slower refresh rate for nearly two years with no problems.
By the way, I'm using "qfresh" from PC Magazine to change the rate.  I've 
forgotten the issue, but can dig it out if anyone is interested.
     To summarize: experiment!
Harold Corwin
-------------
-- 
Harold G. Corwin, Jr.
  UUCP: {backbonesite}!{noao,cs.utexas.edu}!utastro!hgcjr
  Internet: hgcjr@astro.as.utexas.edu           MaBell: 512-471-7463
  Astronomy Dept., RLM 15.308, Univ. of Texas, Austin, TX 78712-1083