[Info-vax] Request for AlphaServer 1000A SIMM/SIMMS

Hans Vlems hvlems at freenet.de
Sat Jul 9 13:21:21 EDT 2011


On Jul 8, 12:04 pm, urbancamo <m... at wickensonline.co.uk> wrote:
> On Jul 8, 7:24 am, Hans Vlems <hvl... at freenet.de> wrote:
>
>
>
>
>
> > On 7 jul, 14:36, urbancamo <m... at wickensonline.co.uk> wrote:
>
> > > Thought I was going mad, but I'm not.
> > > According to the AS1000A Service Guide Bank 0 must contain a memory
> > > option (the lowest bank in the diagram). However, I've found that the
> > > machine won't boot without Bank 3 being populated. Note also that it
> > > specifies five SIMMs  as required. However, when I got the machine it
> > > had Banks 3-1 populated with no SIMMs in the ECC bank.
>
> > > He's the blurb from the manual:
>
> > > 6.2.7 Memory Modules
> > > The positions of the failing single-inline memory modules (SIMMs) are
> > > reported
> > > by SROM power-up scripts (Section 2.2).
> > > Note
> > > • Bank 0 must contain a memory option (5 SIMMs–0, 1, 2, 3, and 1
> > >      ECC SIMM).
> > > • A memory option consists of five SIMMs (0, 1, 2, 3 and 1 ECC SIMM
> > >      for the bank).
> > > • All SIMMs within a bank must be of the same capacity.
>
> > Mark,
> > you wrote earlier that the corresponding ECC slots are empty. I
> > translated that to: they haven't been used by you at any time.
> > If so, what happens if you populate bank 0 and ECO slot 0 and nothing
> > else?
> > There are two reason for this question:
> > - if you're running a system outside its configuration guidelines then
> > expect to see weird behaviour (faultless operation for a system that
> > obviously violates configuration rules is also weird behaviour IMO).
> > - The ECC option is there for a reason, the memory errors reported
> > while booting the system might have been corrected without you ever
> > noticing
> > Hans
>
> My understanding is that with the type-5 system populating the ECC
> slots is not required as the memory does the ECC checking.
> Reading the documentation would suggest that for certain types of
> memory the ECC sticks are different, so I'd need to be sure
> I wasn't making my situation any worse.
>
> Also, the box did come directly from a production environment, so I
> don't expect it was messed about with (the kind of messing about
> that I do for example) before I got my grubby little mitts on it.
>
> Regards, Mark.- Hide quoted text -
>
> - Show quoted text -

OK, that's sound reasoning but then again wouldn't it be nice to know
how the system works when configured according to the good old manual?
You could be living in an area that gets hit by many subatomical
particles (well, whatever) that generate the memory errors you're
getting. :-)
Hans



More information about the Info-vax mailing list