Jim Ablett wrote:EugeneCarey wrote:Jim,
Maybe you can talk Bob Hyatt into running some ratings games on his cluster?
Run these old GNU Chess programs, along with John Stanback's and a few other classics, along with a few reference programs, and see what kind of performance those old things can do.
Shame CrayBlitz isn't Winboard compatable, but I doubt it'd be a heck of a problem to make it work right with Winboard.
Hi Eugene,
Neat idea. Bob would have to recompile the source to his Linux environment though. I don't think he goes anywhere near Windows.
Yeah, his cluster runs a stripped down Linux kernal, if I remember right.
But he has publicly offered to run these kinds of tests, though.
And since these are historical programs rather than somebody's private work, and if CrayBlitz could be a part of it, I doubt he'd decline.
He's pretty much the only one around here than can run this scale of testing in a reasonable time.
I mean, running 6 or 10 or more programs (classics plus reference points) to a couple thousand games.... Not something most people could manage on their home system while they are at work during the day.
I didn't have any trouble compiling Cray Blitz, but coding in Fortran is another story.
I tried auto-converting it to C src using f2c so I could try adding winboard support, but ran into problems.
Let's hope Bob gets around to adding winboard support sometime.
Jim.
First, above, I meant to say "but it'd be a heck of a problem to make it work right with Winboard." but during rewording, I left in the words "I doubt".
Second, it might be feasible to use a translator program to adapt CrayBlitz' I/O to minimal Winboard. But Bob would probably need to work on that.