Sandpiper memory
Dr Ivan D. Reid
going in a skip. Can anyone identify them and suggest if it's worth
trying to sell or give away? (Not much else seemed worthy of rescue; I'd
have had the CPU if I'd had a suitable sized socket spanner with me to
unbolt it.)
Sandpiper MMB Modules 50-21814-01 riser cards, each with four SIMM cards
5021138-01 B1. The intriguing thing is that while each memory card has
that ID, two of the cards in each set are single-sided with ten 44400 chips,
and two are double-sided with 20 44410 chips. I can only assume that the
44410 chips have half the capacity of the 44400s and each board has the
same capacity. I haven't tracked a spec for 44410s, but the 44400s seem
to be 1Mx4 so the cards are presumably each 8 MB ECC (1M x 40).
I'm willing to be corrected on that speculation...
--
Ivan Reid, Electronic & Computer Engineering, ___ CMS Collaboration,
Brunel University. Ivan...@brunel.ac.uk Room 40-1-B12, CERN
KotPT -- "for stupidity above and beyond the call of duty".
Robert Deininger
Ivan...@brunel.ac.uk wrote:
> Just rescued some 1991-era memory modules from a DEC 3000 that was
>going in a skip. Can anyone identify them and suggest if it's worth
>trying to sell or give away? (Not much else seemed worthy of rescue; I'd
>have had the CPU if I'd had a suitable sized socket spanner with me to
>unbolt it.)
>
>Sandpiper MMB Modules 50-21814-01 riser cards, each with four SIMM cards
>5021138-01 B1. The intriguing thing is that while each memory card has
>that ID, two of the cards in each set are single-sided with ten 44400 chips,
>and two are double-sided with 20 44410 chips. I can only assume that the
>44410 chips have half the capacity of the 44400s and each board has the
>same capacity. I haven't tracked a spec for 44410s, but the 44400s seem
>to be 1Mx4 so the cards are presumably each 8 MB ECC (1M x 40).
>
> I'm willing to be corrected on that speculation...
I can't really ID the memory from the chip numbers, but the physical
description should be enough.
The DEC 3000 (sandpiper and flamingo, but not pelican) systems support 6
different SIMM sizes, always installed in kits of 8 matched SIMMs:
1 MB SIMM / 8 MB kit / single-sided / 256k x 4 bit chips
2 MB SIMM / 16 MB kit / double-sided / 256k x 4 bit chips
4 MB SIMM / 32 MB kit / single-sided / 1M x 4 bit chips
8 MB SIMM / 64 MB kit / double-sided / 1M x 4 bit chips
16 MB SIMM / 128 MB kit / single-sided / 4M x 4 bit chips
32 MB SIMM / 256 MB kit / double-sided / 4M x 4 bit chips
The two largest kits are fairly rare because they were so expensive. The
two smallest kits are rare or non-existant because they were too small to
be of much interest. I've never seen definite evidence that DEC ever
released the two smallest kits as products, though the technical
documentation makes it clear that they were part of the system design.
In the console SHOW MEM display, single-sided SIMMs show up as 1 logical
memory bank plus an empty bank. Double-sided SIMMs show up as 2
equal-size logical memory banks. Sandpipers take 2 memory kits (4 logical
banks); flamingos take 4 memory kits (8 logical banks).
A single memory kit is installed across all 4 memory risers. In a
sandpiper, the 8 SIMM slots closest to the motherboard hold one kit, and
the 8 slots farthest from the motherboard hold the other kit.
1M x 4 DRAM chips have 20 pins (4 groups of 5). 4M x 4 chips have 24 pins
(4 groups of 6). I've never seen samples of the smaller SIMMs, so I don't
know what the 256k x 4 chips look like.
DEC and several third parties made memory for these systems, and used
chips from several different vendors. All the modules I've seen match the
description above.
Dr Ivan D. Reid
Robert Deininger <rdein...@mindspringdot.com>
wrote in <rdeininger-21...@user-uinj0cc.dialup.mindspring.com>:
>> Just rescued some 1991-era memory modules from a DEC 3000 that was
>>going in a skip. Can anyone identify them and suggest if it's worth
>>trying to sell or give away? (Not much else seemed worthy of rescue; I'd
>>have had the CPU if I'd had a suitable sized socket spanner with me to
>>unbolt it.)
I did get the CPU in the end; the bolts just holld the heatsink
on the CPU, not also the CPU into the socket...
>>Sandpiper MMB Modules 50-21814-01 riser cards, each with four SIMM cards
>>5021138-01 B1. The intriguing thing is that while each memory card has
>>that ID, two of the cards in each set are single-sided with ten 44400 chips,
>>and two are double-sided with 20 44410 chips. I can only assume that the
>>44410 chips have half the capacity of the 44400s and each board has the
>>same capacity. I haven't tracked a spec for 44410s, but the 44400s seem
>>to be 1Mx4 so the cards are presumably each 8 MB ECC (1M x 40).
I seem to have made a mistake in my arithmetic; that should
have been 4 MB ECC!
>> I'm willing to be corrected on that speculation...
> I can't really ID the memory from the chip numbers, but the physical
> description should be enough.
> The DEC 3000 (sandpiper and flamingo, but not pelican) systems support 6
> different SIMM sizes, always installed in kits of 8 matched SIMMs:
>
> 1 MB SIMM / 8 MB kit / single-sided / 256k x 4 bit chips
> 2 MB SIMM / 16 MB kit / double-sided / 256k x 4 bit chips
> 4 MB SIMM / 32 MB kit / single-sided / 1M x 4 bit chips
OK, that's 4 of them. Still trying to ID the 44410 chips.
> 8 MB SIMM / 64 MB kit / double-sided / 1M x 4 bit chips
> 16 MB SIMM / 128 MB kit / single-sided / 4M x 4 bit chips
> 32 MB SIMM / 256 MB kit / double-sided / 4M x 4 bit chips
> The two largest kits are fairly rare because they were so expensive. The
> two smallest kits are rare or non-existant because they were too small to
> be of much interest. I've never seen definite evidence that DEC ever
> released the two smallest kits as products, though the technical
> documentation makes it clear that they were part of the system design.
> In the console SHOW MEM display, single-sided SIMMs show up as 1 logical
> memory bank plus an empty bank. Double-sided SIMMs show up as 2
> equal-size logical memory banks. Sandpipers take 2 memory kits (4 logical
> banks); flamingos take 4 memory kits (8 logical banks).
> A single memory kit is installed across all 4 memory risers. In a
> sandpiper, the 8 SIMM slots closest to the motherboard hold one kit, and
> the 8 slots farthest from the motherboard hold the other kit.
Check.
> 1M x 4 DRAM chips have 20 pins (4 groups of 5). 4M x 4 chips have 24 pins
> (4 groups of 6). I've never seen samples of the smaller SIMMs, so I don't
> know what the 256k x 4 chips look like.
I'll have to check that next time I'm in the office. I didn't
think to count pins at the time.
> DEC and several third parties made memory for these systems, and used
> chips from several different vendors. All the modules I've seen match the
> description above.
Thanks for the info!
Dr Ivan D. Reid
wrote in <slrnc3fitd.i...@loki.brunel.ac.uk>:
> On Sat, 21 Feb 2004 13:05:45 GMT,
> Robert Deininger <rdein...@mindspringdot.com>
> wrote in <rdeininger-21...@user-uinj0cc.dialup.mindspring.com>:
>> 4 MB SIMM / 32 MB kit / single-sided / 1M x 4 bit chips
> OK, that's 4 of them. Still trying to ID the 44410 chips.
>> 8 MB SIMM / 64 MB kit / double-sided / 1M x 4 bit chips
Looks like that's the other lot.
>> 1M x 4 DRAM chips have 20 pins (4 groups of 5). 4M x 4 chips have 24 pins
>> (4 groups of 6). I've never seen samples of the smaller SIMMs, so I don't
>> know what the 256k x 4 chips look like.
> I'll have to check that next time I'm in the office. I didn't
> think to count pins at the time.
These all have the 20-pin chips, so it looks like I've got 96 MB
of memory up for grabs. Ah, having a closer look, I see why the PCBs have
the same part number -- the 4 MB SIMMs use the same PCB as the 8 MB, but
the second side isn't populated. There are pads there for soldering on
the other 10 memory chips and another component -- probably either a
capacitor or a resistor -- which hides underneath each memory chip.