Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: perft position

Author: Dieter Buerssner

Date: 15:58:51 10/02/04

Go up one level in this thread


On October 01, 2004 at 13:43:17, Uri Blass wrote:

>5b1k/4p1p1/4P1P1/8/8/1p2p1p1/1P2P1P1/K4B2 w - - 0 1
>
>How much time does your program need to get perft 50 and perft 70
>
>latest Movei after fixing perft to be able to calculate more than 20 plies
>gives
>perft 50=531,441
>perft 70=129,140,163

Yace confirms your numbers. Normal perft (which does all the unneeded work, like
updating hashtables; also about 35% of the time is for repetition detection
here, which is automatically called by MakeMove, some more time for this in
UndoMove) needs under 2 s for perft 50, 26 s for perft 60 and 416 s for perft 70
(it calculates all the perft numbers for lower depths on the way):

[...]
48:          354294
49:          531441
50:          531441
51:         1062882
52:         1062882
53:         1594323
54:         1594323
55:         3188646
56:         3188646
57:         4782969
58:         4782969
59:         9565938
60:         9565938
61:        14348907
62:        14348907
63:        28697814
64:        28697814
65:        43046721
66:        43046721
67:        86093442
68:        86093442
69:       129140163
70:       129140163
Used 427.435 s (415.580 s CPU time)

With a hash, it needs practically no time for perft 127:

white ( 1): hperft 127
entries 625000 size 35000000 size_wanted 35000000
127: 1235346792567894
Used 0.000000 s (0.000000 s CPU time)

All on P4, 2.53 GHz.

Can you give an easy formula for the perft numbers of this positition?

Regards,
Dieter



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.