Writing the Assembler

[Gary Wheeler]

See attached WORD file

[Bob Flanders]

Thank you for this, Gary.

On Wed, Jul 31, 2024 at 12:53 PM Gary Wheeler <garypw...@hotmail.com> wrote:

See attached WORD file

--
IBM1130 discord channel: https://discord.gg/nrFEMt48mE
---
You received this message because you are subscribed to the Google Groups "IBM1130" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ibm1130+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ibm1130/PH0PR08MB72587C20567CF49942397FB3D6B12%40PH0PR08MB7258.namprd08.prod.outlook.com.

[John Pierce]

On Wed, Jul 31, 2024 at 9:53 AM Gary Wheeler <garypw...@hotmail.com> wrote:

See attached WORD file

For those who don't/won't open .DOC files...  here's the plain text version...

When I joined IBM in 1963 as a new college graduate from Stanford, I was joining as the only

programmer in a group responsible for the software of what was to become IBM’s Small

Engineering Computer (aka 1130). I had essentially no programming experience save for two

summer internships with IBM. In fact, at that time there was no Computer Science department

at Stanford. The others in the department had no computer experience. They were all ex-

salesmen whose background was selling IBM punched card equipment. As the only

programmer, I was told that my assignment was to write the software for the new system

(FORTRAN, Assembler, utilities, and some type of supervisor). With this background, I

basically spent the first six months not knowing where to start. Finally, IBM transferred an

experienced (and very bright) programmer from Endicott, NY, and now I had a mentor.

Meanwhile development of the new system was proceeding, but with complication. As was

often the case at the time, there was another team in Poughkeepsie, NY pushing competitive

architecture. For a time, when their design had more favor, we were looking at what I think was

an 18-bit architecture. Then San Jose would gain favor and we’d go back to 16-bit architecture.

As you know, the 1130 did not have floating point arithmetic hardware, so one of the metrics

used to evaluate the competitive designs was to code routines for floating point arithmetic and

evaluate how much storage was used and what the operation execution time. My job at this

point was to code the floating-point routines for the San Jose design and a counterpart in

Poughkeepsie would code the routines using their design.

Finally, the decision was made for the San Jose design, and I began working on a bootstrap

assembler that team (now including a group developing FORTRAN) could use to build the code

to be tested on an actual 1130. The bootstrap assembler was written for an IBM 1401,

assembled punched card segments of the software and punched out program decks to be

loaded and executed on the 1130.

Before starting on the Assembler, it had to be decided what the language would look like, what

opcodes and pseudo-opcodes would be used. Since it would use punch card input, the layout

of a card had to be chosen (length of labels, format indicator, designation of index register, etc.)

As the Assembler design proceeded it was decided that it was to be a two-pass assembler and

what tasks would be accomplished in each pass.

Challenges:

1. The paper tape only version (literally rolls of paper tape to be read each time a test case

was run)

2. Debugging with a printout of core storage

After my work on DM2 I moved to Boca Raton to manage the Test and Release group.

Returning to California following three years in Boca Raton to coordinate the release of 1130

DSP and the incorporation of other languages such as RPG, COBOL, and APL. All in all, I was

involved with the 1130 for the first ten years of my time at IBM. It was a great experience! 

[Paul Anagnostopoulos]

Super, thanks!

~~ Paul

[Paul Anagnostopoulos]

I really enjoyed reading your comments about your time in the 1130 software group. Thanks again.

Can you recall any interesting details about coding the assembler? In particular, how difficult was it to cram it into 4K?