What AWIT(y) chess program. Tony Marsland's, that is....
Posted: Wed Mar 17, 2010 3:47 am
This is just to let people know that Tony Marsland has released the source for his selective search chess program AWIT from the 70's and 80's.
For those of you who don't remember Tony Marsland, he was one of the people who organized the first US computer chess tournament. He also did a lot of computer chess research in the 70s and 80s.
The program is written in ALGOL-W, a predecessor of Pascal. At the time that was a very reasonable choice, but as time went on, the language & compiler became more of a problem. Limitations of the language forced him to write some less than pretty code. Limitations of the compiler forced even more complicated code in an effort to get better speed. Eventually, the program became so big that the one existing Algol-W compiler couldn't even compile it without special effort to trick it.
It should have been rewritten, but he never got around to it. He was busy doing research with other programs, had limited mainframe time, was busy with life in general, etc.
So this is still the AlgolW code that was used and experimented with for so many years.
Good for history, not so good for those of us who would like to really run the program to best of its abilities on modern hardware. (Yes, I admit it... I would like to see how well a selective search program works. I know the math behind why selective search is considered risky, but look at all the stuff modern programs do now... I've never been a fan of brute force searching. It just seems.... wrong.)
With the help of Glyn Webster's AW2C converter program, Mr. marsland was able to get the program running again.
Unfortunately, with the speed of modern hardware, we were able to play more self play games in a few days than Awit had played in its entire life. And that let a few bugs show up.
Unfortunately, this version was the only version of Awit that still existed. He didn't have any ealier versions to see when or where the bugs might have been written. And trying to debug a program you hadn't looked at in 25 years is a bit of a challenge.
But with some persistence, he was able to get things working fairly well.
And its finally ready to be released.... So if you'd like to see what a classic selective search program looks like:
http://webdocs.cs.ualberta.ca/~tony/Pub ... uterChess/
Mr. marsland is also looking for any photos of him running Awit, so he could add them to his website.
For those of you who don't remember Tony Marsland, he was one of the people who organized the first US computer chess tournament. He also did a lot of computer chess research in the 70s and 80s.
The program is written in ALGOL-W, a predecessor of Pascal. At the time that was a very reasonable choice, but as time went on, the language & compiler became more of a problem. Limitations of the language forced him to write some less than pretty code. Limitations of the compiler forced even more complicated code in an effort to get better speed. Eventually, the program became so big that the one existing Algol-W compiler couldn't even compile it without special effort to trick it.
It should have been rewritten, but he never got around to it. He was busy doing research with other programs, had limited mainframe time, was busy with life in general, etc.
So this is still the AlgolW code that was used and experimented with for so many years.
Good for history, not so good for those of us who would like to really run the program to best of its abilities on modern hardware. (Yes, I admit it... I would like to see how well a selective search program works. I know the math behind why selective search is considered risky, but look at all the stuff modern programs do now... I've never been a fan of brute force searching. It just seems.... wrong.)
With the help of Glyn Webster's AW2C converter program, Mr. marsland was able to get the program running again.
Unfortunately, with the speed of modern hardware, we were able to play more self play games in a few days than Awit had played in its entire life. And that let a few bugs show up.
Unfortunately, this version was the only version of Awit that still existed. He didn't have any ealier versions to see when or where the bugs might have been written. And trying to debug a program you hadn't looked at in 25 years is a bit of a challenge.
But with some persistence, he was able to get things working fairly well.
And its finally ready to be released.... So if you'd like to see what a classic selective search program looks like:
http://webdocs.cs.ualberta.ca/~tony/Pub ... uterChess/
Mr. marsland is also looking for any photos of him running Awit, so he could add them to his website.
