Jay West wrote: Going by the card list presented by the seller and what can be discerned from the picture of the card cage, the unit is missing the core memory boards. Other memory module boards for what should be 16-24 KWords are present but the actual core boards appear to have been pulled. (There are also some discrepancies between the card list and the card cage picture.)

Brent wrote... I didn't look that close. Brent is of course right, memory seems to be missing. Something else caught my eye upon re-visiting that ad. Either that machine never ran HP2000 TSB, or someone has pulled a lot more cards than just the memory. I'm quite certain that a system with those cards could not run TSB as is. My guess is either someone pulled lots of I/O cards, or, the seller did a quick google search on 2116 and found some mention of TSB so he assumed it was a TSB box. Not necessarily at all.

>Wasn't there a TSB which used two 2116's, one for computation > and one for I/O?

I recall configurations that used a 2116 with a 2114 as the I/O processor; I never actually saw a 2116/2116 combination.

I had written.... To which Brent replied.... Well, I am *very* familiar with the hardware requirements for TSB, and I'm obviously insane. Ok, this is especially embarassing - since HP2000 TSB is my specialty and also the centerpiece of my collection. OBVIOUSLY senility has set in, probably coupled with the fact that I have focused on restoring the DEC pieces in my collection as of late and demonstrably have forgotten everything about my 3 beloved HP TSB machines. Perhaps it's time for an exorcism, to have my DEC daemons cast out so I can get back to my beloved HP2100 machines :) Brent is again completely correct - this does seem to be a valid configuration for the I/O processor in several different versions of TSB. The only thing amiss to me is Brent said there were two GND TRU boards, but I only see one listed in the auction. You would have to have two for sure as the IPCK was always four 12566 boards with 2 on each end. One pair used for sending and one pair used for receiving. Virtually all TSB variants required 2 cpus. 2000B, 2000C, 2000C', 2000F, 2000F', and 2000/Access all required two cpus. 2000A and 2000E required only one cpu. There are docs for each version stating which processors are valid for the main cpu and which are valid for the IO cpu. I have long suspected that those specifications are incorrect, and that the main and I/O cpu's can be any combination. While C, E, F, and Access OS's are around - 2000A and B don't seem to be so I'm not sure. I would wager that 2000/Access makes the most demands of the IOP due to the IBM/CDC MRJE stuff, and it's requirements for memory in the IOP are anywhere from 16kw to 32kw depending on software options selected at IOP generation time. I would think that the prior versions of TSB would certainly not need more than Access so I'd have to suspect the memory indicated by the non-SSA cards in the memory subsystem is adequate. If someone needs them let me know. I have a pretty huge number of spare HP2100/21MX cards, including the SSA boards that are used by the 2116C. In addition the SSA boards are generally pretty easy to repair. Ok, I'm going to go sit in the corner and hide now! Jay

The only thing amiss to me is Brent said there were two GND TRU boards, but I only see one listed in the auction. You would have to have two for sure as the IPCK was always four 12566 boards with 2 on each end. One pair used for sending and one pair used for receiving.

I would wager that 2000/Access makes the most demands of the IOP due to the IBM/CDC MRJE stuff, and it's requirements for memory in the IOP are anywhere from 16kw to 32kw depending on software options selected at IOP generation time. I would think that the prior versions of TSB would certainly not need more than Access so I'd have to suspect the memory indicated by the non-SSA cards in the memory subsystem is adequate.