Thanks for the pointer to memtest86. I downloaded a copy after findi=
ng it=20
at freshmeat.net, burned it to CDROM and booted the system in question fr=
om=20
the CDROM.
     The problem follows the smaller (128M) RAM stick but the failure onl=
y=20
shows up when both sticks are present: tested one-at-a-time, everything w=
orks=20
but with both present, the failure shows up in the address range of the=20
smaller stick.
      Wierd, but conclusive enough for me.
      Thanks for the solution!



On Sunday 22 June 2003 16:02, Deepak Saxena wrote:
> On Jun 22 2003, at 15:55, Ed Skinner was caught saying:
> > Hmmm,
> >      I have a x86 motherboard (Giga-byte GA-7VKMLS, in another Walmar=
t
> > box) that doesn't seem to mind two differently sized RAM sticks in A/=
B
> > order but becomes intermittent (POST doesn't start 50% of the time at
> > power on) when in B/A order. That is, if I swap the sticks, it works,=
 and
> > if I swap them back, it is intermittent. The MOBO book says mixing si=
zes
> > is Okay but obviously something is going on.
> >      The two sticks are marked:
> > Value RAM, KVR-PC133/256, 3.3v, Kingston
> > Value RAM, KVR133x64c3/128, 3.3v, Kingston
> >      Anyone recognize if I'm doing a No-No
>
> Run memtest86 on the RAM.  It's possible that something is wrong with
> the low-address range stick B and it's only getting hit during POST.
>
> ~Deepak

--=20
Ed Skinner, ed@flat5.net, http://www.flat5.net/