Re: DS5000 memory

[Henry Bent <hbent%oberlin.edu@localhost>]

On 3 May 2010 00:23, Ahnjoan Amous <ahnjoan%gmail.com@localhost> wrote:
> Just checking out a number of memory modules I've purchased over the
> years using "test" at boot prompt and recieved the following.
>
>>>test
> ?TFL:  3/mem (2: board 5: too many SBEs: 4194304) [KN03-AA]
> ?TFL:  3/mem (1: board 7, MBE= 8388608, SBE= 0) [KN03-AA]
> ?TFL:  3/mem (2: board 10: too many SBEs: 4194304) [KN03-AA]
> ?TFL:  3/ni/ext-lb (3:(xmt [00000006])) [KN05]
> ?RTCsi/cntl
> ?TFL:  3/mem (2: board 5: too many SBEs: 4194304) [KN03-AA]
> ?TFL:  3/mem (1: board 7, MBE= 8388608, SBE= 0) [KN03-AA]
> ?TFL:  3/mem (2: board 10: too many SBEs: 4194304) [KN03-AA]
> ?TFL:  3/ni/ext-lb (3:(xmt [00000006])) [KN05]
> ?RTC
>>>
>
> So I moved the board in slot 5 to slot slot 8, slot 7 to slot 10, and
> slot 10 to slot 9.  This in an effort to identify the problem memory
> modules.  I then restarted the DS and ran test again with results that
> indicate the problem doesn't follow the memory shift.  The following
> output is from another issuance of "test".
>
> If anyone has an idea of what these messages mean I would appreciate it.
>
>>>test
> ?TFL:  3/mem (2: board 4: too many SBEs: 31948) [KN03-AA]
> ?TFL:  3/ni/ext-lb (3:(xmt [00000006])) [KN05]
> ?TFL:  3/mem (2: board 4: too many SBEs: 31948) [KN03-AA]
> ?TFL:  3/ni/ext-lb (3:(xmt [00000006])) [KN05]
>
> Thanks
> Ahnjoan

Hi,

SBEs are single bit errors.  They are the sort of errors that ECC
memory, the kind used in DECstations, is designed to identify and
correct.  If, however, there are too many single bit errors it
generally indicates a problem with the memory.  The first step is to
thoroughly clean all of the contacts on the memory and the board; I
generally use the somewhat low-tech approach of some rubbing alcohol
and and old toothbrush.  You can also try a pencil eraser on the
memory module contacts if they have become oxidized.  If that fails I
would suspect that the main board is failing and needs to be replaced.
 In my experience this sort of thing is increasingly common with pmax
machines; in the last four years I have had three machines (two /120s
and one /25) fail in essentially unfixable ways.

-Henry Bent