LM wrote:
> Has anyone tried adding second gigabyte to this motherboard. I am
> using XP SP3. This PC is now quite stable with 1 G ram, but when put
> in second gig, all sorts of problems start to appear. Like problems
> with Firefox browser, random bluescreens, problems after standby and
> so on. This is now second configuration I tried with same results.
> Memtest seems to be ok. And I'm using the machine now, so it almost
> works. I don't think any drivers are needed, but something else is
> wrong here.
>
> Have you seen this? Comments?
>
Get a copy of memtest86+ (
http://www.memtest.org/ ).
Try testing the sticks one at a time first, and
do a couple complete test passes. That will
give you a quick answer, if one of the sticks
is bad.
Memtest86+ is available for floppy diskette or CDROM.
For the CD, you use a tool that knows how to burn
an ISO9660 file to the CD. (Nero knows how to read
a ISO9660 and make a bootable CD from it. Don't just
copy the file straight onto the CD.)
I've used 2x512MB and 4x512MB on mine, without
a problem. The extra loading of four sticks, is
only an issue if you're overclocking the memory.
The chipset should easily be able to run 4x512MB
at DDR400 without a problem. But you cannot get
to an overclock like DDR500 with four sticks. So
the extra bus load prevents overclocking to the
same degree, as you can with two sticks. The record
for the P4C800-E is somewhere around DDR600 with
a lightly loaded bus.
You can experiment with extra Vdimm voltage. The
JEDEC committee thinks 2.6V is enough for DDR400
memory operation. I've used 2.65V or 2.7V on some
boards here without a problem. Not all RAM is rated
for high voltage. The record holder is stuff like
the older Winbond memory chips, and some people
ran those at 3.3V or more. But more conventional
memory, like some Micron chips, might die if you
try something like that. 2.7V should be enough
to aid stability a bit.
Only use as much voltage as is needed to get
stability and error free operation. If you see
only a small number of random errors without
Vdimm boosting, then chances are good a little more
voltage will fix it. If you're getting hundreds of
thousands of errors, then the stick is likely
busted. Similarly, if the same location gives errors
every time, that could be a bad chip. I've had
one stick of memory here, where a single bit
was stuck at one value and could not be
changed.
When all your testing is finished, then you can
try booting an OS. I usually run a bootable Linux
CD for a while first, before I go back to Windows.
Linux is my favorite way to test hardware, but I
don't use it as a daily OS. The Linux CD cannot be
corrupted by bad memory, whereas your Windows
boot drive can be corrupted.
Paul