Monday, July 22, 2019

"The Real Story Behind the Apollo 11 Computer Error"; WSJ short film is pertinent now, 50 years later

I’ve presented some New York Times short films (including Oculus made with the help of Annapurna) but here is a short film from the Wall Street Journal today, “The Real Story Behind the Apollo 11 Computer Error”.

The narration concerns computer programmer Don Eyles, my age, who walked into an MIT lab in 1966 (at the same time I had started graduate school at KU) and asked for a job with no coding experience.
He wound up writing the memory-efficient “lapidary” assembler or machine code that the lunar module would use.  It was fit into six sequential pieces of “core rope”, each 36K, presumably each with its own register (rather like base register concepts in IBM mainframe assembler code which I worked on in the 1980s and 1990s and which the IRS still uses a lot of today – I know that from oral job interviews in the 2000s;  coding “out of addressability” was a good trick).  I actually wrote some similar shipboard code for NAVCOSSACT (Washington Navy Yard) in 1972, but never implemented it, as I took a job with Univac, and office mates went down to Charleston SC and boarded a ship later to implement it.  I had actually learned FORTRAN code in summer jobs at the David Taylor Model Basin (Maryland, on the DC Beltway) in three summers starting in 1965 and small computers existed then.
He says he almost became a life insurance salesman (like an agent?)  That’s ironic.

The core ropes were among the first use of integrated circuits in computers.  Up to that time, IBM (and to some extent Sperry Univac and Burroughs) had dominated the mainframe business market with systems that filled whole computer rooms (like the 1110 benchmark I worked on in St Paul MN in the spring of 1974).

The failure was caused by signals from nearby hardware with a switch that malfunctioned.
Eyles says this was a sickening moment, just three years after starting work. But Mission Control in Houston figured out that his code still worked and the problem was external and that the astronauts would be able to land.  The rest is history.

No comments: