Author: Vincent Diepeveen
Date: 10:43:12 04/07/04
Go up one level in this thread
On April 07, 2004 at 12:36:38, Renze Steenhuisen wrote: >On April 07, 2004 at 12:16:05, Vincent Diepeveen wrote: > >>On April 07, 2004 at 11:40:36, Renze Steenhuisen wrote: >> >>> >>>I understand your frustration completely! I am CS myself in Delft (as you might >>>remember) and know what you talk of... >>> >>>>Do you actually know what boundschecking means? >>> >>>However, in Delft we have (since last year) a course called Software Quality and >>>Testing of which I was the teaching assistant. So yes I know very well what >>>boundschecking is and means. And I have seen many students that didn't >>>understand what it was, or at least didn't grasp how to do it correctly! >>> >>>I could show you some code of those guys if you like, it is very shamefull! >>> >>>Cheers! >> >>I hope you give that to students in their first months when they start the >>course. >> >>Now using this course you should be able to do some debugging in your program >>and bugfix it all. > >Yeah, but there are still A LOT of lines and checks to do. And I simply haven't >got the time the check it all right now... I use perft() for checking the >move-generator, do_move and undo_move on 4 positions to different depths so it >is crunching for 30 minutes or so. All tests pass. > >I think I just tested the incremental hashkey as well. I have put a permanent >check in for returning to the original state, from which we were searching. > >For now this is just part of my graduation project. It does not need to be >perfect and completely bug-free and monkey proof! It should be monkey proof. A bug in your hashtables will lead you to conclude all kind of weird things. What you can do is not do too much effort to write very fast code. Usual scientific parallel running software is already a factor 30 slower anyway (Zugzwang, Cilkchess, P.Conners), so if you manage to reduce it to a factor 10 and then get a speedup > 10 at 64 nodes that would be very cool. Your only effort should be to get at 64 nodes a speed equal to what i get at a dual k7. Best regards, Vincent
This page took 0.01 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.