Author: john tromp
Date: 09:13:17 12/06/04
Go up one level in this thread
On December 04, 2004 at 18:04:56, martin fierz wrote: >>However, I found that the program still lacked in strength, since >>it failed to see that some new threats would inevitably arise >>that would change its perceived outcome. This turned out practically >>impossible to correct, so I gave up on that approach. > >i wonder: i probably need some time to update threats etc that you don't need, >but on the other hand, having an idea where the game is going should improve my >move ordering. do you think it's not worth it? my experience with history >heuristic in connect 4 is terrible to say the least, i'm not using it at all, it >always made things worse for me. Then what do you use for move ordering? A static evaluator? Is the source to your engine available, either publicly, or privately? >my current program solves connect 4 in 1520s on an AMD64 3000+ with 128MB >hashtable. it searches 4.55 GN to do this. how does your program do? Please download and run the Fhourstones benchmark and let me know:-! I hope your AMD64 is fast enough to beat the current record holding overclocked Pentium-M (4102 Kpos/s) http://www.cwi.nl/~tromp/c4/fhour.html For my desktop machine, the result with 48Mb hashtables is Solving 0-ply position after . . . 3215994182 pos / 1548481 msec = 2076.9 Kpos/sec which is quite comparable to your searchsize. It could also solve it with 640Kb of hashtable but that will take at least twice as long. regards, -John
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.