Author: martin fierz
Date: 07:56:56 10/29/03
Go up one level in this thread
On October 29, 2003 at 09:54:59, Steven Edwards wrote:
>On October 29, 2003 at 09:48:13, Robert Hyatt wrote:
>>On October 28, 2003 at 23:46:56, Steven Edwards wrote:
>>>On October 28, 2003 at 23:37:28, Robert Hyatt wrote:
>>>>On October 28, 2003 at 21:10:37, Steven Edwards wrote:
>
>>>>>Here's a one question historical quiz that will test your knowledge of our
>>>>>little corner of the programming world:
>>>>>
>>>>>Who was/were the first programmer(s) to use bitboards for piece
>>>>>location/property representation?
>>>
>>>>without reading, I believe was both Slate/Atkin in chess 4.0 in 1974, and
>>>>the russian group working on Kaissa, with Donskoy as the main programmer.
>>>
>>>Of course, both the NWU Chess and the Kaissa programs used bitboards, and in
>>>apparantly an identical manner. Possibly Kaissa also used them in its causality
>>>facility ("method of analogies").
>>>
>>>>They apparently discovered this idea independently, at about the same time,
>>>>and both showed up with bitboard programs at the same time...
>>>
>>>True. But neither is the answer to the question! I know you know the answer,
>>>so please give it another try.
>>
>>
>>OK... perhaps you intended a bid of misleading here? IE I assumed "chess". If
>>you are not specifically talking about chess, then my second choice would be
>>Samuel's checker program. He didn't need 64 bit words of course, since checkers
>>only has 32 usable squares.
>
>Yes, Arthur Samuel was the first. Here's the rot-13 of the rot-13:
>
>The answer is not slate and atkin with their nwu chess program nor is
>it donsky et all from moscow with their kaissa program. The first
>bitboard program was the famous checker player for the ibm mainframes
>from the fifties written by arthur samuel. Born in the first year of
>the twentieth century he was at the time of his passing eighty nine
>years later probably the world's oldest active computer programmer.
>His pioneering work was influencial in having the various boolean
>bitwise operators being standard in modern instruction sets, without
>which today's chess bitboard programs would be infeasible. His papers
>described how he came up with the idea of embedding a thirty two bit
>representation of the active squares on a checker board in the thirty
>six bit word of the ibm mainframes he had available. His program also
>used machine learning, an opening book, transposition detection, and
>many other techniques all long before their appearences in chess
>playing programs.
damn, i should have known :-)
as a checkers programmer, i read the article of samuel of course, a long time
ago (that's the only excuse i have - except that i also didn't notice that the
question was not only about chess programs).
i have one more thing to add: this "strange" 36-bit word of the ibm mainframes
(at least for me it seems strange) is better suited to representing the checker
board than the modern 32-bit word, because he could choose his bit <-> board
representation so that (bitboard>>C1) would always be a left-forward move, and
(bitboard>>C2) a right forward move. replacing >> with << goes backward of
course. with just 32 bits you can't do that, and it's a bit more complicated for
us today (only marginally, but still).
cheers
martin
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.