Computer Chess Club Archives


Search

Terms

Messages

Subject: Illegal EPD file in Neishtadt

Author: Vincent Diepeveen

Date: 22:37:13 11/21/02

Go up one level in this thread


On November 22, 2002 at 01:14:51, Vincent Diepeveen wrote:

position 40 in your testset is illegal:

4kqrk/r5pp/pn1p1p2/2pPpPPQ/2P4P/1P3P2/2BBK3/6RR w - - 0 1 [280]

I wonder how you ran tests with it.

>Hello, first table is first part of Omids numbers he needs at
>
>http://www.cs.biu.ac.il/~davoudo/pubs/vrfd_neishtadt_dep9.txt
>
>Node counts to get a 9 ply search depth.
>
>        std R = 1       std R = 2       std R = 3      vrfd R = 3
>
>  1       6864391         2073995          353415          640836
>  2      11849513         3952080          704417         1159792
>  3       7735232         2421290         1024175         1909431
>  4      10399311         3217656          296633          424892
>  5       7894973         1667965          391265         1276789
>  6      23843259         8257609         9196191         7599069
>  7      16628446         4849724          774861         1275569
>  8        232149          100757           25096           47493
>  9       7923159         2472308          282501          542888
> 10       8422519         2967494          277978          519170
> 11       5882451         2211559          418959          630473
> 12      16328807        10896537         3521012         3423127
> 13      11757759         3931197         1670286         4404093
> 14      52221384        15187127         7874393        13271141
> 15      25414826         8068577          802341         1448106
> 16       4928787         1841152          744449          986228
> 17      29153860         8656690         7331342        19243173
> 18      19251137         6351804         5102125         7498985
> 19       2676023          981646          416481          636074
> 20      36591070        14202065         6044385         8603409
> 21      22981115        16732702        17323332        13160168
> 22       1850548          662302          334199          748723
> 23       1329803          499360          213462          350438
> 24      25077330         7829115         1872259         3658545
> 25       7918374         3624752         2546515         6415498
> 26      12932272         3954053         1515639         2121015
> 27       7918018         2639096          765158         1687489
> 28       7931073         2880664         1432676         2767042
> 29       3068185         1122273          146023          281129
> 30      10010362         2893682         1196519         2498546
> 31       1785611          622960          111294          161102
> 32      21267135         7950051         3624877         8643335
> 33       4507858         2653606         1259310         2984433
> 34       5462440         1951888          413763          756776
> 35      15510610         6237615         2680171         5205710
> 36      19031416         8431949         6134553        10015012
> 37      23244769        10064705         4283335         2791961
> 38       3982956         2258114         1285883         1117010
> 39        540635          226331           90965          146517
> 40       7366081         3515047         1956719          686776
> 41       4465020         1442182          489696         1152792
> 42      10442881         5683547         1495948         4652067
> 43       1698094          623525           95375          244035
> 44        560198          436933          335003          167386
> 45      27686232         8698978         4577878         9033014
> 46       4688588         1802221          583648         1320446
> 47        921944          282597          100937          224661
> 48       4140231         1368459          321342          707243
> 49      13663065         4713142         2904570         6137891
> 50       9705458         3848893         1637528         2530890
>
>
>
>From Omid's positions the vaste majority is mating positions.
>
>position 4 is a mate in 2. How do you manage to need 10.4MLN nodes
>there Omid to finish 9 ply.
>
>Here is what DIEP needs at 9 ply for the
>same positions and fullwidth it ain't much more for the mate in 2:
>
>depth totalnodes              time needed (single cpu)
> 9  223019 (      0,      0)    3.02  1
> 9   54038 (      0,      0)    0.83  2
> 9 1707338 (      0,      0)   21.74  3
> 9    9814 (      0,      0)    0.27  4
> 9  145529 (      0,      0)    1.98  5
> 9 1343273 (      0,      0)   19.77  6
> 9  142710 (      0,      0)    1.34  7
> 9   37725 (      0,      0)    0.81  8
> 9  178479 (      0,      0)    2.05  9
> 9   11620 (      0,      0)    0.44  10
> 9   53039 (      0,      0)    0.89  11
> 9 2348386 (      0,      0)   25.78  12
> 9 1541525 (      0,      0)   18.14  13
> 9 5177632 (      0,      0)   63.75  14
> 9  120584 (      0,      0)    1.57  15
> 9   43201 (      0,      0)    0.76  16
> 9 3599264 (      0,      0)   49.54  17
> 9 1253990 (      0,      0)   16.47  18
> 9  173255 (      0,      0)    2.95  19
> 9  186748 (      0,      0)    1.56  20
> 9 3836215 (      0,      0)   53.99  21
> 9  183931 (      0,      0)    1.67  22
>
>So you need a factor 1000 more with R=1 there
>than DIEP needs with nullmove.
>
>296613 nodes for R=3 is also a weird high number
>for a mate in 2, considering how simple your qsearch
>seems to be.
>
>Not a single position is not getting solved.
>Not a single position therefore we have a regular
>search seraching the next ply. It's just fail high fail high
>fail high.
>
>And majority is mates. Perhaps it's all mates even...
>
>How can you need a 6.9 branching factor with R=1 when it is
>basically all mates?
>
>10 million nodes for a mate in 2 to finish 9 ply!!!!!!!!!!!!!!!!!!!!!
>
>This is weird.
>
>Best Regards,
>Vincent



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.