Author: Ross Boyd
Date: 14:23:15 03/15/05
Go up one level in this thread
On March 15, 2005 at 08:54:22, Joel wrote: >I have been working in fits and starts. I think I started the new version >September/October last year? Something like that. Currently working hard at the >moment! Good to hear. I knew you'd be up to mischief. > >I have the new Bodo working on ICC with Polyglot. Everything seems to work >perfectly - PolyGlot must like my UCI implementation more than Arena. I have >tried analysis mode, and played a few games on ICC with no problems. Thanks >Fabien! Working already!? Great! Yep, WB seems more reliable for playing online. > >I still have no book code - I was hoping to have it playing using an Arena made >book initially. Apparently I can use BookThinker (whatever that is) in >conjunction with PolyGlot...something for another day. > >Yes, I hard to swallow my pride and chuck out a lot of my old code. I decided >that I had made a mess, largely since I didn't know what was involved in making >a chess engine when I started previously. The old engine had major evaluation >bugs, and poor time management - I think these were the two main factors holding >it back. Nod. I think most first engines end up messy. Its too hard to resist the temptation to just get something working. And it takes some experience before the right structure emerges.... unless of course your name is Fabien or Perola etc... > >New Bodo: >C++ (C with classes really) versus C this time round. >Better design internally (not hard compared to last time). >It is still BitBoard, not out of preference this time, but because my old move >generator was correct, and I didn't see enough gain to throw it away. I just >C++'ified it. >Emphasis on simplicity >and the biggest reason for the rewrite, I am preparing to use more than one cpu. Hey, that sounds ominous! I wouldn't have a clue how to implement a parallel engine. >I will compile it in 64bit mode soon, maybe a couple of weeks. It isn't >optimised at all, so I won't bother until it is. >It is still weaker than the old bodo, but it won't be long before it catches up. >If you get an ICC account, you can watch its progress live! :) I _hope_ (and >expect) it to be stronger than the old bodo in a month - I see promising signs >now. We shall see, we shall see!! :-) >I changed protocol since UCI seemed to be less work for what I want to do with >the engine. With PolyGlot, I have more working Winboard features than I did >previously! :) That's bizarre. Fabien is one smart cookie. He writes the most elegant code. I have tried to adopt a similar style in TRACE 2. Truckloads of asserts, and help the optimiser by declaring pointers and data const wherever possible. Maybe this year at NC3 we can give Warp and co a real run for their money. :-) Keep up the good work! Ross
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.