Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Bad PV's

Author: Dann Corbit

Date: 17:04:53 07/16/04

Go up one level in this thread


On July 16, 2004 at 20:01:24, Dann Corbit wrote:

>On July 16, 2004 at 19:49:37, Stuart Cracraft wrote:
>
>>Hi -- I have spent days on the hashing and
>>am ready for something else for awhile.
>>Found 4 or 5 hashing bugs so far. A real
>>hornet's nest that one...
>>
>>18 ASSERT's in overall code are showing that measurables
>>being measured are all okay and not violating the
>>ASSERT's. Thank you Bob Floyd, R.I.P., 1st programming class.
>>
>>On the other hand, my PV per iteration has gone
>>to hell in a handbag. Anyone have a "usual suspect"
>>for the dogmove 2. ... d7d5 below at 4th iteration?
>>My max positional (mp) sums to typicaly < 1/5 of a pawn
>>so it's not that.
>>
>>When the ASSERT's were violating before the
>>hash work, the PV's were always relatively clean.
>>
>>1. e2e4  0.00   31       21 e2e4
>>2. e2e4  0.00    0       87 e2e4 e7e5
>>3. e2e4  0.01   17      276 e2e4 e7e5 f1c4
>>4. e2e4  0.04    0     1714 e2e4 e7e5 f1c4 d7d5
>>5. e2e4  0.12   17     7636 e2e4 e7e5 g1f3 d7d5 e4d5
>>6. e2e4  0.43    0    31106 e2e4 e7e5 f1c4 d7d5 e4d5 d8d5
>>7. e2e4  1.57   17   121299 e2e4 e7e5 f1c4 d7d5 e4d5 d8d5 c4d5
>>8. e2e4  5.64    0   433830 e2e4 e7e5 f1c4 d7d5 e4d5 d8d5 c4d5 b8c6
>>e2e4  5.64    0   433830 e2e4 e7e5 f1c4 d7d5 e4d5 d8d5 c4d5 b8c6
>>nps=76892 ha=29.00% q=81.0% bc=72% br=3.6% mp=178<>178
>>pawnx=0 recapx=0 qcheckx=25250 checkx=0 futilx=0 qfutilx=40132
>
>The pawn move is bad, but it is a lot less suspect than tossing the queen!
>Perhaps your evaluation is inverted for black?  He looks like he's in a hurry to
>lose.

[D]rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq -
[D]rnbqkbnr/pppppppp/8/8/4P3/8/PPPP1PPP/RNBQKBNR b KQkq -
[D]rnbqkbnr/pppp1ppp/8/4p3/4P3/8/PPPP1PPP/RNBQKBNR w KQkq -
[D]rnbqkbnr/pppp1ppp/8/4p3/2B1P3/8/PPPP1PPP/RNBQK1NR b KQkq -
[D]rnbqkbnr/ppp2ppp/8/3pp3/2B1P3/8/PPPP1PPP/RNBQK1NR w KQkq -
[D]rnbqkbnr/ppp2ppp/8/3Pp3/2B5/8/PPPP1PPP/RNBQK1NR b KQkq -
[D]rnb1kbnr/ppp2ppp/8/3qp3/2B5/8/PPPP1PPP/RNBQK1NR w KQkq -
[D]rnb1kbnr/ppp2ppp/8/3Bp3/8/8/PPPP1PPP/RNBQK1NR b KQkq -



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.