The Electronic Coach
April 10, 2012 7:09 PM Subscribe
In the main link in griphus' post this morning, there was this little aside: "In 1957...a physics student named Don Knuth built a program for the IBM 650 to help the 1958 Case Institute of Technology basketball team win the league championship."
Yes, THAT Don Knuth. Here's a young Don with the team and the IBM 650 (capable of making 50,000 calculations a minute!), and here he is talking about it.
Details in chapter 23 of Selected Papers on Fun and Games
Don Knuth previously 1, 2, etc.
Details in chapter 23 of Selected Papers on Fun and Games
Don Knuth previously 1, 2, etc.
My favorite part of that video is how much taller Knuth is than the rest of the basketball team.
posted by DU at 7:26 PM on April 10, 2012 [2 favorites]
posted by DU at 7:26 PM on April 10, 2012 [2 favorites]
Yeah, no offense to Don Bluth, but there's at least an order of magnitude of difference between his importance and Don Knuth's.
posted by sonic meat machine at 7:41 PM on April 10, 2012
posted by sonic meat machine at 7:41 PM on April 10, 2012
50,000 calculations a minute is 833 hertz. Not kilohertz. Not megahertz. Certainly not the modern gigahertz. Just, plain, hertz. Operations per second.
By way of comparision, the Apple II was clocked at one megahertz, a million instructions per second. And it had (up to) 64K of RAM, where that IBM had 2000 memory slots, each ten digits long. If we assume that those were hexadecimal digits (which may not be accurate) then each number would have been a nybble, half a byte. So that first machine kinda-sorta had about the equivalent of 10K of RAM, but addressed very strangely. Worse, it wasn't even really RAM... it was apparently semi-permanent storage, on a metal drum. That's part of why it was so incredibly slow. I guess they hadn't invented RAM yet, so the machine would have been doing a hard drive seek for each and every byte of data.
In other words, those guys would have been willing to shoot people to get their hands on an Apple II. :-)
posted by Malor at 7:51 PM on April 10, 2012 [2 favorites]
By way of comparision, the Apple II was clocked at one megahertz, a million instructions per second. And it had (up to) 64K of RAM, where that IBM had 2000 memory slots, each ten digits long. If we assume that those were hexadecimal digits (which may not be accurate) then each number would have been a nybble, half a byte. So that first machine kinda-sorta had about the equivalent of 10K of RAM, but addressed very strangely. Worse, it wasn't even really RAM... it was apparently semi-permanent storage, on a metal drum. That's part of why it was so incredibly slow. I guess they hadn't invented RAM yet, so the machine would have been doing a hard drive seek for each and every byte of data.
In other words, those guys would have been willing to shoot people to get their hands on an Apple II. :-)
posted by Malor at 7:51 PM on April 10, 2012 [2 favorites]
Nerds should never help jocks. Ever.
posted by symbioid at 7:51 PM on April 10, 2012 [3 favorites]
posted by symbioid at 7:51 PM on April 10, 2012 [3 favorites]
I love this little factoid:
'Since version 3, TeX has used an idiosyncratic version numbering system, where updates have been indicated by adding an extra digit at the end of the decimal, so that the version number asymptotically approaches π... The current version of TeX is 3.1415926... [Knuth] has stated that the "absolutely final change (to be made after my death)" will be to change the version number to π, at which point all remaining bugs will become features.'
posted by urbanwhaleshark at 7:56 PM on April 10, 2012 [5 favorites]
'Since version 3, TeX has used an idiosyncratic version numbering system, where updates have been indicated by adding an extra digit at the end of the decimal, so that the version number asymptotically approaches π... The current version of TeX is 3.1415926... [Knuth] has stated that the "absolutely final change (to be made after my death)" will be to change the version number to π, at which point all remaining bugs will become features.'
posted by urbanwhaleshark at 7:56 PM on April 10, 2012 [5 favorites]
Core memory existed in the fifties, but it was ridiculously labor-intensive to make: each bit had to be wound by hand. According to Wikipedia, IBM introduced an optional magnetic-core storage unit for the 650 in 1955, capable of storing 600 digits. I would love to know how much it cost, but that information doesn't seem to be online.
Of course, no thread about drum memory computers is complete without a link to The Story of Mel.
posted by teraflop at 8:18 PM on April 10, 2012 [3 favorites]
Of course, no thread about drum memory computers is complete without a link to The Story of Mel.
posted by teraflop at 8:18 PM on April 10, 2012 [3 favorites]
833 hertz
Slightly more than a G sharp. Not quite an A flat.
posted by ceribus peribus at 8:21 PM on April 10, 2012 [6 favorites]
Slightly more than a G sharp. Not quite an A flat.
posted by ceribus peribus at 8:21 PM on April 10, 2012 [6 favorites]
50,000 calculations a minute is 833 hertz.
According to Bob LaFara, the 650 had a clock cycle time of 100usec, and a basic integer add took 8 cycles. This works out to 75,000 adds per minute.
If you want to try turning the knobs yourself, there's a 650 simulator. (with manual, describing the instruction set and other details)
posted by MtDewd at 9:10 PM on April 10, 2012 [1 favorite]
According to Bob LaFara, the 650 had a clock cycle time of 100usec, and a basic integer add took 8 cycles. This works out to 75,000 adds per minute.
If you want to try turning the knobs yourself, there's a 650 simulator. (with manual, describing the instruction set and other details)
posted by MtDewd at 9:10 PM on April 10, 2012 [1 favorite]
I loved how the Grantland guy dismissed Knuth as just some computer dude. And by loved I mean I almost came out of my seat with a savage,
posted by ob1quixote at 3:21 AM on April 11, 2012 [2 favorites]
You're talking about Donald Knuth, you philistine!
posted by ob1quixote at 3:21 AM on April 11, 2012 [2 favorites]
This works out to 75,000 adds per minute.
Right, though in order to go from clock speed to "instructions per second", a weighted scale was needed to balance out the different speeds & complexities of the calculation. (division takes longer than addition, etc.) So they came up with an approximation of 50,000 instructions, assuming a certain mixture of different operations.
posted by ShutterBun at 9:41 AM on April 11, 2012
Right, though in order to go from clock speed to "instructions per second", a weighted scale was needed to balance out the different speeds & complexities of the calculation. (division takes longer than addition, etc.) So they came up with an approximation of 50,000 instructions, assuming a certain mixture of different operations.
posted by ShutterBun at 9:41 AM on April 11, 2012
Slightly more than a G sharp. Not quite an A flat.
The difference between a G# and an A♭ being what, exactly?
posted by ShutterBun at 9:47 AM on April 11, 2012
The difference between a G# and an A♭ being what, exactly?
posted by ShutterBun at 9:47 AM on April 11, 2012
« Older Walter Hill's "The Warriors" | "Don't turn it on - take it apahhht!" Newer »
This thread has been archived and is closed to new comments
posted by ancillary at 7:18 PM on April 10, 2012 [1 favorite]