🚀 go-pugleaf

In article <830872851.28840.0@nook.demon.co.uk> you write:

| I am looking for some way to check that memory is functioning 100%
| correctly.  I have run Norton Utilities (under DOS) memory check facility,
| and this reports no errors, but this check runs under DOS.
|
| My machine is recently prone to mysterious hangs, crashes & GPFs in Win3.11,
| not just under certain circumstances, but at random.
|
| I recently added 8MB memory to the configuration, to give me 16MB in all.
| The original 8MB is W/Parity 70ns, and has 10 chips on each side of the
| simm.  The new simm has chips on only one side, 2 sets of 3, with a smaller
| one in the middle.  It is also W/Parity 70ns.

  I'd worry about the 'smaller one'.  If its really quite small it sounds
  as though you have fake parity.  Instead of providing memory to store
  the parity bit you have a chip that calculates it when you read the SIMM.
  Of course this is _useless_ since you cannot compare what you wrote
  with what you are reading.  I'd query this with the supplier.

  It's also possible that your original memory is _better_ than 70ns and
  you have setup your motherboard to work with it.  It maybe that you
  need to go into your BIOS setup and add a wait state.

|
| Ok, so I would normally source additional memory from the same supplier as
| the original, but a. the original supplier doesn't deal in hardware any
| more, and b. the price was pretty good.
|
| So --
|
| 1.  Am I just stupid to use memory that has a different external appearance
| (though the vital specs are the same)?

   No absolutely not.  There are almost as many ways of skinning a cat
   as there are manufacturers.  I think you could almost expect them to
   be different.

|
| 2.  Are there any Public Domain tools which will run under Windows to test
| the memory exhaustively?
|

   I very much doubt it. Running under Windows would render the test
   _completely useless_.  The whole point about Windows (well any OS of
   that type; I'm not slagging off Windows though I do happen to hate
   it) is that it stops programs having any control over where in memory
   they are and _prevents_ them accessing memory they shouldn't.  That's
   what a GPF is about.  If you don't have absolute access to all known
   memory you simply cannot test it.

|
| Thanks for any help, suggestions, comments etc.
|
   Note that running 'under' DOS is a good way to test memory  'cos you
   can access almost all of the memory directly. Under is a misnomer in
   this case.  A program running in a bare m/c can acess all the memory
   both above and below the 1MB boundary. Even better would be to boot
   off a floppy with DOS low and no EMM386 in the config.

   Having said all that I don't know of any good memory test programs.
   There are some in the Simtel archives under memutil but they all date
   from the stone age (pre 286?) and don't seem to work on the later
   m/c's.  I would have thought this was an active area for PD writers
   and/or Motorola/NEC/Hitachi/Samsung  would all have complimentary
   test programs out there to promote their image and their product.
--
Peter Hayward, Cambridge, UK.      | In a democracy the public will only
peter@cdsl.demon.co.uk             | respect laws if they respect the
tel +44-1223-312-923               | politicians that make them.