Newsgroups: comp.sys.transputer
From: paul@walker.demon.co.uk (Paul Walker)
Subject: Re: inmos IMS 2620P-12 IC's, what are these?
Organization: Paul Walker Consultancy
Date: Wed, 16 Mar 1994 07:40:05 +0000
Message-ID: <763803605snz@walker.demon.co.uk>

In article <CMpDG3.89y@ndl.co.uk> davidb@ndl.co.uk "David Boreham" writes:

>In article <2m1eb7$ejk@thor.cs.utexas.edu>, cpg@cs.utexas.edu (Carlos M.
> Puchol) says:
>>
>>I have a bunch of integrated circuits from an old (I guess) transputer
>>board for a PC. These chips have the following on top of them:
>>
>>inmos 8506-B
>>IMS 2620P-12
>>
>>Can someone tell me (through email preferably) what kind of chips are
>>these? I would guess they are memories, but they may not be.
>
>Its a 64K DRAM, 16K by 4. 120ns access time.
>
>The other number is the date code, indicating that the device was tested
>and marked in the sixth week of 1985.
>
>For those interested in trivia, the numbering scheme works like this:
>
>The first digit signifies the device type. "1" is SRAM, "2" is DRAM, 3 is
> "EEPROM"
>The second digit is the number of bits, as in 2 to the power of the (digit +10)
>The third digit is the word width, as 2 to the power of the digit.
>The last digit is for function variations, as in Nibble mode etc.
>
>(Is that right Paul ?)

Absolutely right David, although the second digit might be better 
understood as log [base 2] of the number of K of the device. The 
numbering scheme ran out of steam about the time INMOS stopped 
making DRAMs, when the next number had to be IMS 2A00 (1 Meg x 1).

The real expert on all this is David Wooten, now at Compaq.

Could also try Rahul Sud, at SGS-THOMSON Agrate.
>
>I can't even remember which boards had these in, must have been a B001 or
>something totally old. Perhaps the first B004s had these in ?
>If you need to replace these devices, look for other manufacturers devices,
>probably numbered "4416" or something like that. These parts are long
>obsolete though. Your best bet for a datasheet is to get hold of an old memory
>databook for any DRAM manufacturer cira 1985/6 and look for the 16K by 4s,
>they call used the same pinout.


Not B001, because that was SRAM;
Not B002, because that used 256K x 1s to give 2 Mbytes
Not B004, ditto (check with andyr@wizzy.com who designed the B004?)
The B003 used 4-bit wide DRAMs, but I thought was 256KBytes per transputer.

And I think the B000 (never sold, and with the S42 processor rather than a 
real transputer) used SRAM as well.

So were these 2620s ever used on a transputer board?

Best regards

Paul Walker
Consultancy in product specification and design
(and apparently in the history of INMOS hardware!)

