Re: [vcf-midatlantic] Looking for some knowledgeable with the Harvard Mark I
There are some glorious YouTube videos of Grace Hopper showing the wide paper-tape containing the programming, and how she had to literally patch it (cover holes, add holes, etc). I vaguely remember her mentioning how it was an early form of parallel processing: calculations could be run simultaneously on each functional unit (so long as there were no data hazards). Such optimization was meticulously planned by hand. In a way, we're revisiting that with super-scalar RISC, DSPs or EPIC (Explicitly Parallel Instruction Computing). My point being there's WHAT it can do (instruction set, data paths) and then there's HOW it was used. The compelling historical element is how it was all PEOPLE, not software. But Grace Hopper's Flow-Matic was an amazing leap forward. Like Alan Turing, she understood that computers can process SYMBOLS and handle all the drudgery of programming! -- jeff jonas
On Sunday, October 11, 2020, 1:40:30 AM EDT, Jeffrey Jonas via vcf-midatlantic <vcf-midatlantic@lists.vcfed.org> wrote:
There are some glorious YouTube videos of Grace Hopper showing the wide paper-tape containing the programming, and how she had to literally patch it (cover holes, add holes, etc).
Very cool! I'll have to look those up after VCF.
I vaguely remember her mentioning how it was an early form of parallel processing: calculations could be run simultaneously on each functional unit
Yes. In the early documentation, they refer to it as interposition. The manual has a wonderful quote: "in order to attain the maximum speed of computation, full advantage must be taken of the methods of interposition." The Mark I really was quite an interesting machine. BLS
participants (2)
-
Brian L. Stuart -
Jeffrey Jonas