On 5 Oct 99, at 20:32, Kyle Elmblade wrote:

> Actually, I would like to confirm this issue.  I haven't tested
> this on Windows 98, but I have on Windows 95.  I used a resource
> monitor (it was about 6 months ago, and I can't remember which one
> now) that looked specifically at the memory usage.  I had a machine
> with 96MB of RAM installed, and Windows 95 reported all 96MB in the
> system properties.  So, I started opening up web pages and watching
> the system monitor during each instance.  As soon as I hit 60MB of
> RAM (I had the memory cache in my browser set to 96MB) I started
> getting memory/resources low warnings from Windows.  As soon as I
> hit 64MB, the system started freezing and I had to start shutting
> things down, eventually having to reboot the system entirely.
>
> Next I did the same thing by opening a text file I had created to
> be exactly 1024K bytes.  I created 64 copies and started opening
> them one by one.  At about the 55th file I started getting
> memory/resource warnings again.  When I hit the 64th file, I had to
> again start closing files to get windows to respond properly.

  Resources are a special kind of entity that must live within a
relatively small area; the size of this area doesn't change when you
add/remove RAM.  It could be important to understand whether these
were memory or resource warnings.

> I then removed 32MB of RAM to make a system with 64MB.  The system
> ran, based on various speed test, an average of 20% faster.  Keep
> in mind, this is all with Windows 95B OSR 2.1.  To me, this meant
> that Windows 95 can only use up to 64MB of RAM, and above that
> actually detracts from the efficiency of the system.  I would be
> real interested to hear if anyone else has run any tests like
> these, and what their results were.

  The worst penalty for installing more than 64MB of RAM on a system
with this limitation that I've seen documented previously was 15% on
one particular motherboard model using a 430TX chipset (which has the
same limitation).  Interestingly, the same source showed another
430TX-based board where the penalty was only 3%.  Apparently not all
430TX boards were created equal in this regard, and the same may be
true of 430VX boards.

> ....  I haven't tested
> this on Windows 98, but I have on Windows 95.

  It is my understanding that Win98 is supposed to make better use of
larger RAM configurations than Win 95 did.  I'd be very interested in
seeing this experiment repeated....

David G

         The PCBUILD web site always needs good submissions.  If
          you would like to contribute to the website, send any
               hardware tech tips or hardware reviews to:
                           [log in to unmask]