Call for manuals and maybe floppies: IBM 8100

[Jay Jaeger]

On 8/26/2021 7:54 PM, Paul Berger via cctalk wrote:
> 
> On 2021-08-26 6:48 p.m., Jay Jaeger via cctech wrote:
>> My next project once I finish my IBM 1410 FPGA implementation (so, a 
>> couple of years out, probably) would be to write an emulator for the 
>> boat anchor known as the IBM 8100.  I had exposure to these things 
>> back in the 1980s.  The project was not really a success: the DPPX 
>> operating system was way overkill for the underpowered machine, and 
>> wasn't reliable enough or capable enough to run them at remote 
>> locations with central administration.
>>
>> The machine had some fairly sophisticated features:
>>    Two groups of 64 sets of registers with 8 32 bit registers each
>>    Auto increment and auto decrement indexed addressing
>>    Address translation - but not paging
>>    A primitive form of I/O channel
>>
>>
>> True story:  The early releases of DPPX were just awful buggy.  We 
>> ended up dedicating 3 conference rooms (with the dividers open) for a 
>> "warm room" for something like 3 months, housing our personnel and IBM 
>> personnel up from Texas.  At one point one of the IBM'ers was 
>> overheard on a public phone in the hallway of our public building 
>> telling someone he was there "to help the hicks from Wisconsin".  That 
>> got reported to our management and to IBM's management, and he was on 
>> the next flight back to Texas.  ;)
>>
>> On the flip side, I was testing database recovery (it was my thing, 
>> back in the day - though we did not end up using the database / 
>> transaction manager).  I found some bugs in the database log journal 
>> recovery process.  I mentioned it to one of the IBM'ers in passing, 
>> also pointing out it wasn't urgent since we were not going to use DTMS 
>> anyway, at least not soon.  He pretty much begged me to report it - 
>> and anything else I found wrong.  Completely polar opposite attitude 
>> of the guy in the previous paragraph.
>>
>> JRJ
> 
> The 8100 came out during my first stint in field service, most of the 
> machines we saw where 8130s that ran the DPCX operating system and they 
> where purchased to replace 3790 distributed processors. We had one 
> customer who bought an 8140 and I helped with a model upgrade on it, 
> which involved removing all the logic gates as one unit and replacing 
> them with the upgraded one.  It would have been quicker and probably 
> cheaper to ship an entire new machine, but they where advertised as 
> field upgradeable so.....   A few years later when I saw the inside of a 
> S/38 I recognized the packaging of the system as being identical to the 
> 8140 except the 8140 did not have the built in CRT console or magazine 
> diskette drive. This 8140 was running DPPX but I don't know how the 
> customer got on with it as it was not my account.  It is said that the 
> 8100 processor is the Universal Controller (UC), but I am hoping it was 
> a beefed up on from the UC engine that ran several of the "Industry 
> Systems" controllers such as 3274 (NDS), 3601 (Banking), and 3651 
> (retail store systems).
> 
> Paul.
> 
> 

I had a look at the UC information on 
bitsavers.org/pdf/IBM/microcontrollers, and also looked at some field 
service material available on archive.org.

The CPU comprises several boards, so it wasn't UC per se.  There was a 
ROS-driven micro controller that handled instruction fetch and 
branching, but it had 56 bit words per that field service material.

It may have had a similar software architecture to the UCs, but at 
present from what I have seen the similarity seems to end their.

JRJ