DIMM Question

[Guest]

Archived from groups: alt.sys.pc-clone.gateway2000 (More info?)

I have a Gateway 710x which I just upgraded by adding 2-512mb DIMMs.
Unfortunately, I didn't get one of the DIMMs properly seated and got some
very unexpected results since the system specified that they be installed in
pairs. When I powered it up only one of the DIMMs was recognized by the
BIOS; Memtest 86 also told me I only had one memory chip and then went ahead
and checked it with all tests passed.

Can someone, with more background than I, tell me why it worked with the
unbalanced chip install (3 working & one missing)? I would have expected a
failure at the very least since I understood that the bus would assign one
memory location on one chip and the next on the other which in this case was
absent.

Thanks for any info you may have,
Tom

 

[Guest]

Archived from groups: alt.sys.pc-clone.gateway2000 (More info?)

"Tom Clydesdale" <t.clydesdale.nospam@comcast.net> wrote in message
news:s-SdnZyhZqcF-3DcRVn-gA@comcast.com...
>I have a Gateway 710x which I just upgraded by adding 2-512mb DIMMs.
>Unfortunately, I didn't get one of the DIMMs properly seated and got some
>very unexpected results since the system specified that they be installed
>in pairs. When I powered it up only one of the DIMMs was recognized by the
>BIOS; Memtest 86 also told me I only had one memory chip and then went
>ahead and checked it with all tests passed.
>
> Can someone, with more background than I, tell me why it worked with the
> unbalanced chip install (3 working & one missing)? I would have expected
> a failure at the very least since I understood that the bus would assign
> one memory location on one chip and the next on the other which in this
> case was absent.
>
> Thanks for any info you may have,
> Tom
>

Tom,

Dual-channel DDR does not require all DIMM sockets to be populated to
function. The situation you describe should result in only one of the new
DIMMs being recognized, or at the very worst the system might revert to
single-channel mode while detecting all DIMMs properly installed.

This in comparision to the old RDRAM/Rambus systems which required *all* (4)
sockets on both channels to be populated and in matched pairs to properly
POST and function.

As an experiment, you can simply place one DIMM in one socket of your system
(leaving the others out) and the machine will POST and boot in
single-channel mode - a performance hit, but still functional.

I'm not sure what happens on a single pair of mismatched DIMMs on the same
channel, as I haven't had the time to blow that one up yet

hth

Stew