Not logged inRybka Chess Community Forum
Up Topic The Rybka Lounge / Computer Chess / Toga CMLX 1.4 beta 5c!
- - By Vempele (Silver) Date 2008-02-17 07:25
Enjoy. I don't have any time right now, more info will follow later.
Attachment: Toga960.zip (209k)
Parent - - By RedGambit Date 2008-02-17 09:51
is this engine stronger than toga 1.4.1SE or toga 1.4.1Beta5c ?
Parent - - By Vempele (Silver) Date 2008-02-17 14:49 Edited 2008-02-17 14:52
Considering neither of those play Chess960, it definitely is stronger. Apart from Chess960, it´s identical to Toga II 1.4 beta5c.

Edit: it´s somewhat slower, though. I barely had time to compile it.

I´m in NORWAY? Here I thought Copenhagen Airport was part of Denmark; it certainly was four days ago!
Parent - - By Vempele (Silver) Date 2008-02-17 18:46 Edited 2008-02-17 19:04
Reminds me of when the forum said I was in both Serbia and Montenegro at the same time. They'd been separate countries for more than a year. :)
Parent - - By Vempele (Silver) Date 2008-02-17 18:49
...
Parent - - By noctiferus (***) Date 2008-02-17 18:59
What a traveller: faster than the light.. :)
Parent - By Uly (Gold) Date 2008-02-17 19:18
Maybe he's using a rotating IP proxy.
Parent - - By FWCC (***) Date 2008-03-05 20:43
This version only uses two cores?I see only two used in my quad.
Parent - By Vempele (Silver) Date 2008-03-05 21:29
http://mihd.net/i5hlqn builds by Denis Mendoza. Much faster than mine, too.

It'll be configurable in 5e.
Parent - - By NATIONAL12 (Gold) Date 2008-02-17 22:37
i am testing Naum 3-toga 11 1.4.1SE at the moment at 14/40  2 core 1.83 GHz hs book 12 moves 50 games 256 hash shold have full details within 48 hours. 32 bit. i hope this will help various toga developers.
Parent - - By NATIONAL12 (Gold) Date 2008-02-19 13:11
Naum 3-Toga 11 1.4.1 SE. +18=20-12. 40/14. 2 core,HS masterbook.12 moves.256 hash.32 bit. 1.83Ghz laptop. a good result, the strongest TOGA i have tested.
Parent - - By NATIONAL12 (Gold) Date 2008-03-02 22:23
i have justfinished testing Toga CMLX 1.4 beta5d-Naum 3 under the following conditions,40/40,win32,quad 2.4GHz,HS masterbook 12moves,512 hash. +10=18-12. this incredible result for a free engine. incidentally i cannot reproduce on a core duo at 1.83GHz at time contol 40/14 where Naum 3 is a big winner.this approx 7 times slower than my quad result. i have started testing against Mexico 11 so far results confirm Togas strength, +1=4-1.
Parent - By NATIONAL12 (Gold) Date 2008-03-02 23:13
could i ask a question of the 5d developer. what do you estimate the speed up between 2 core and 4 core  on win 32 at long time controles or do you rely on testing similar to mine,from my own testing toga is much better at tournement time controls.this is on win 32 which i believe the great unwashed as turbo would put it have.
Parent - - By NATIONAL12 (Gold) Date 2008-03-05 21:53
finished testing Toga CMLX 1.4 beta5d-Mexico 11 at 40/40 conditions as above,win 32 quad.  +6=26-8,very very good,it would have been even better but toga with B&NvK could not find the winning line. started testing against Rybka.
Parent - - By Eelco de Groot (***) Date 2008-03-06 22:46 Edited 2008-03-06 23:01
Thanks for the testing National, it is quite a coincidence that Bishop and Knight vs. King alone came on the board in your games just now :). I would have thought the engame is pretty rare, but that is maybe from the perspective of human games where most players don't want to test their opponents knowledge of this endgame, and least not GMs. There was a discussion on the CSS forum that was the reason for me to look at the code again and it was mentioned there that in his whole career Victor Korchnoj maybe never got the chance to play this endgame, not one game in a database of more than 7000 games! And yet in a small match between two computers the endgame comes on the board after just a few games.

It is a small piece of code and most was already written by Thomas Gaksch, maybe it is an idea to put this small routine in the next CMLX 5e version or in Thomas' latest -6a?- version? I have to admit I would like to borrow some of Teemu's code for Chess960 in time, I don't think I could write such a thing myself, unless I study other programs with FRC code,  that of course would help. The way I implemented KBNK is also heavily based on Tord's code in Glaurung although Tord's version is more compact and he wrote a version that computes things just in one piece of code for the needed colour.

Here is my version of the KBNK routine in eval.cpp:

static const int KBNKingposMatrix[64] = {
      90,    70,    50,    30,    10,   -10,   -30,   -50,
      70,    40,    20,     0,   -20,   -40,   -60,   -30,
      50,    20,   -10,   -30,   -50,   -70,   -40,   -10,
      30,     0,   -30,   -60,   -80,   -50,   -20,    10,
      10,   -20,   -50,   -80,   -60,   -30,     0,    30,
     -10,   -40,   -70,   -50,   -30,   -10,    20,    50,
     -30,   -60,   -40,   -20,     0,    20,    40,    50,
     -50,   -30,   -10,    10,    30,    50,    70,    90,
};

.

static const int KingDistanceBonus[8] = {0, 0, 50, 40, 30, 20, 10, 5};

.
.

// eval()

int eval(board_t * board, int alpha, int beta, int ThreadId) {

   int opening, endgame;
   material_info_t mat_info[1];
   pawn_info_t pawn_info[1];
   int mul[ColourNb];
   int phase;
   int eval;
   int wb, bb;
   int winnerKSQ, loserKSQ;
   int lazy_eval,probe_score; // Thomas

   ASSERT(board!=NULL);
   ASSERT(board_is_legal(board));
   ASSERT(!board_is_check(board)); // exceptions are extremely rare

   // init

   opening = 0;
   endgame = 0;

/*   if (egbb_is_loaded && board->piece_nb <=5 && ABS(SearchBest[0].value) >= WIN_SCORE) {
    if (probe_bitbases(board, probe_score)){
      return probe_score;
    }
   } */

   // material

   material_get_info(mat_info,board,ThreadId);

   opening += mat_info->opening;
   endgame += mat_info->endgame;

   mul[White] = mat_info->mul[White];
   mul[Black] = mat_info->mul[Black];

   // PST

   opening += board->opening;
   endgame += board->endgame;

   // pawns

   pawn_get_info(pawn_info,board,ThreadId);

   opening += pawn_info->opening;
   endgame += pawn_info->endgame;

   // draw

   eval_draw(board,mat_info,pawn_info,mul);

   if (mat_info->mul[White] < mul[White]) mul[White] = mat_info->mul[White];
   if (mat_info->mul[Black] < mul[Black]) mul[Black] = mat_info->mul[Black];

   if (mul[White] == 0 && mul[Black] == 0) return ValueDraw;

    // win

   if (board->piece_nb == 4){ // KBNK?
      if (board->number[WhiteKnight12] == 1 && board->number[WhiteBishop12] == 1){
         eval = ValueWin;
         wb = board->piece[White][1];
         ASSERT(PIECE_IS_BISHOP(board->square[wb]));
         winnerKSQ = KING_POS(board,White);
         loserKSQ = KING_POS(board,Black);
         if (SQUARE_COLOUR(wb) == SQUARE_COLOUR(A8)){
            winnerKSQ = ((winnerKSQ)^0x0F);
            loserKSQ = ((loserKSQ)^0x0F);
         }
         eval += KBNKingposMatrix[SquareTo64[loserKSQ]];
         eval += KBNKingposMatrix[SquareTo64[loserKSQ]];
         eval += KingDistanceBonus[DISTANCE(KING_POS(board,White),KING_POS(board,Black))];
      }
      else{ // KKBN?    
             if (board->number[BlackKnight12] == 1 && board->number[BlackBishop12] == 1){
                eval = ValueWin;
                 bb = board->piece[Black][1];
                 ASSERT(PIECE_IS_BISHOP(board->square[bb]));
                 winnerKSQ = KING_POS(board,Black);
                 loserKSQ = KING_POS(board,White);
                 if (SQUARE_COLOUR(bb) == SQUARE_COLOUR(A8)){
                    winnerKSQ = ((winnerKSQ)^0x0F);
                    loserKSQ = ((loserKSQ)^0x0F);
                 }
                 eval += KBNKingposMatrix[SquareTo64[loserKSQ]];
                 eval += KBNKingposMatrix[SquareTo64[loserKSQ]];
                 eval += KingDistanceBonus[DISTANCE(KING_POS(board,White),KING_POS(board,Black))];
             }
      }
      if (COLOUR_IS_WHITE(board->turn)){
         return (eval);
      }
      else{
             return (-eval);
      }
   } // End of KBNK block
.
.
.


As you can see it is a small piece of code, if the values in the KBNKingposmatrix are doubled you don't need to call the piece square matrix twice but that was just for tuning purposes not changed yet. Maybe some finetuning is possible by using the KBNK pst value of the winning King too, a small bonus for occupying the centre as well as advancing towards the enemy King may work especially if all pieces somehow ended up clustered in the 'wrong' corner where mate is not possible, in that case the white King should just back off a little first with his troops and occupy the centre, at least I think that is the human theory.
Parent - - By NATIONAL12 (Gold) Date 2008-03-06 22:56 Edited 2008-03-06 23:16
i just test the engines for my own pleasure, i am not a programer but i will always post interesting things like this. about a week ago i was playing a toga engine against rybka again at 40/40 and rybka gave a +6 to itself and toga went p queens=N dead draw. its also nice to know that someone reads my posts as i am completly independant.
Parent - - By Ty Nance (**) Date 2008-03-06 23:38 Edited 2008-03-06 23:41
is "queens" the verb you want, when pawn promotes to knight? More seriously, as I've read the discussion about the newest, strongest, fastest Toga, I think I want it for myself. I don't care about Chess960, so I don't need that option if it slows the program at all. Where can I find and download the best Toga as a UCI engine?

Thanks,

ty
Parent - By NATIONAL12 (Gold) Date 2008-03-07 00:10
yeh ok my grammar is not that good at 12pm uk. i cannot remember off hand where i got the download from but dont worry just ask vempole as he is the compiler of 5d and is on this forum,he corrected the errors in 5c.
Parent - - By Vempele (Silver) Date 2008-03-07 10:53

> It is a small piece of code and most was already written by Thomas Gaksch, maybe it is an idea to put this small routine in the next CMLX 5e version or in Thomas' latest -6a?- version?


I've adapted it now. Expect 5e later today.

> especially if all pieces somehow ended up clustered in the 'wrong' corner where mate is not possible, in that case the white King should just back off a little first with his troops and occupy the centre, at least I think that is the human theory.


No. It's like this (some of black's moves are suboptimal):

5k2/5N2/5K2/5B2/8/8/8/8 w - - 0 1


1. Bh7 Ke8 2. Ne5 Kd8 3. Ke6 Kc7 4. Nd7 Kc6 5. Bd3 Kc7 6. Bb5 Kd8 7. Kd6 Ke8
8. Bc4 Kd8 9. Bf7 Kc8 10. Nc5 Kd8 11. Nb7+ Kc8 12. Kc6 Kb8 13. Be6 Ka7 14.
Nc5 Kb8 15. Kb6 Ka8 16. Bf5 Kb8 17. Na6+ Ka8 18. Be4#
Parent - By Eelco de Groot (***) Date 2008-03-07 11:18
That looks really nice Teemu! Good news, I'm glad to hear that Toga CMLX beta5e is almost finished!

Regards, Eelco
Parent - - By NATIONAL12 (Gold) Date 2008-03-07 21:48
are we still going to get 5e in next 24 hours.
Parent - - By Vempele (Silver) Date 2008-03-07 22:06
Yes. And it'll almost certainly still be today as constrained by my previous post, at least in some parts of the world. :)

It'll have played a few hundred games by then.
Parent - - By NATIONAL12 (Gold) Date 2008-03-07 22:08
what time controls do you test at.
Parent - - By Vempele (Silver) Date 2008-03-07 22:09
60 moves/minute.
Parent - By NATIONAL12 (Gold) Date 2008-03-07 22:27
christ i test at 40/40.
Parent - By Vempele (Silver) Date 2008-03-08 07:44
It crashed, and I only got 84 games. Plus I forgot Glaurung defaults to using both cores.
Parent - By Vempele (Silver) Date 2008-03-08 15:43

> Expect 5e later today.


Or not. I think I'm finally close to getting the original TogaMP fixed. It's based on 1.2.1 so I'm going to port some changes first. SMP efficiency is going to jump. :)
Parent - - By Vempele (Silver) Date 2008-02-18 09:35 Edited 2008-02-18 09:40
Just found the bug that caused crashes for Denis. My fault for never testing with non-mirrored kings.

New version to come soon - I'll fix some other bugs (not mine).
Parent - - By Ty Nance (**) Date 2008-03-07 00:25
Vempele:

I asked National 12, and he said to ask you:

As I've read the discussion about the newest, strongest, fastest Toga, I think I want it for myself. I don't care about Chess960, so I don't need that option if it slows the program at all. Where can I find and download the best Toga as a UCI engine?

thanks,

ty
Parent - By Eelco de Groot (***) Date 2008-03-07 10:02 Edited 2008-03-07 10:30
Just have some patience till 5e is ready. There are several links to 5d mentioned in this thread. (It is not the Chess960 options slowing it down but better to leave off any bitbase pruning in the second version of Toga CMLX, version 5d, because that does slow the engine and the bitbase probing is not functioning.) If you don't care about Chess960 there are numerous other versions you could try. It is not clear which of these is the strongest, it also depends on your system and what kind of timecontrols you prefer. In the CCRL list version 6a from Thomas is doing well but that is also a private version. If you already have 5c I suggest you just use that one. There are many versions of Toga all with their own flavour, just try some I would say :) Regards, Eelco
Parent - - By Alexander Schmidt Date 2008-03-02 15:59
Thanks for teaching Toga Chess960 :)

Unfortunately I cant get it work under Arena with Polyglot... Toga accepts castling, but Polyglot don`t accept Togas castlemoves. Any ideas?

Alex
Parent - - By Vempele (Silver) Date 2008-03-02 16:52
Fuck Arena! Way not to implement FRC:

-ignore the standard
-be anal about O-O vs 0-0
-don't bother to document it

Extra points if your FENs are broken.

In other words, I neglected to test it under Arena because three quarters of the above brokenness disgusted me too much.

Wait for the next version. You're not supposed to need Polyglot, by the way.
Parent - By Alexander Schmidt Date 2008-03-02 17:39
Hi,

just one thing: Arena had the first UCI/Chess960 protocol, so you can not blame Arena for ignoring the standard :)

Polyglot translates Shredder-UCI-Chess960 in WB-Chess960, so far all Shredder-UCI-Chess960 engines worked also with Polyglot under Arena.

BTW: I guess it is not necessary to support Arena-Chess960-UCI, I am not sure if future Arena releases support it, and in the current Arena beta it is broken.

Alex
Parent - - By BB (****) Date 2008-03-02 22:52

> !@$#%^ Arena!


Should the typical forum courtesy also apply to GUIs? :)
Parent - - By Vempele (Silver) Date 2008-03-03 13:38 Edited 2008-03-03 14:29

>> :-( %/^@#!


>Should the typical forum courtesy also apply to GUIs?


FYT (fixed your typo, not the other meaning). :-p
Parent - - By BB (****) Date 2008-03-03 14:14
If this is going to be the typical forum language, I shall cease from posting in the future.
Parent - By Uly (Gold) Date 2008-03-03 23:08

> If this is going to be the typical forum language, I shall cease from posting in the future.


Don't worry, Vempele has been the only member not censored around, so this language happens rarely.
Parent - - By Eelco de Groot (***) Date 2008-03-05 01:52
About the bitbase probing, typical scores are or at least were around 40+. This exceeds ValueMate in Toga CMLX which I think could be a problem? I increased ValueMate again to 10000 (100 pawns) and in the KBNK routine it seems to help a little against phoney matescores that I got. I still have to figure out why I have to switch comparing the color of the Bishop square with h8 instead of a8, it is a complete mystery to me but hopefully not related to the mate scores...

With ValueMate = 4000 (40 eval pawns) I got things like this, without bitbase probing switched on;


7N/6K1/6B1/8/8/8/8/7k w - -


Engine: Toga 1.5 Checkov_beta1 bb5men (64 MB)
by Letouzey, Gaksch, Pudas and de Groot

1/01   0:00   +31.90   1.Bb1 (2)

1/02   0:00   +32.00   1.Kf6 (13)

2/06   0:00   +31.10   1.Kf6 Kg2 (35)

2/06   0:00   +31.60   1.Be4+ Kg1 2.Kf6 (146)

3/06   0:00   +30.70   1.Be4+ Kg1 2.Kf6 Kf2 (204)

3/06   0:00   +31.20   1.Kf6 Kg2 2.Ke5 (235)

4/06   0:00   +30.80   1.Kf6 Kg2 2.Be4+ Kf2 3.Ke5 (645)

5/08   0:00   +30.80   1.Kf6 Kg2 2.Bh5 Kf2 3.Ke5 (1.582)

6/08   0:00   +29.90   1.Kf6 Kg2 2.Bh5 Kf2 3.Ke5 Ke3 (3.315)

6/14   0:00   +30.10   1.Nf7 Kg2 2.Ne5 Kf2 3.Nc4 Kf3 (8.680)

7/14   0:00   +30.20   1.Nf7 Kg2 2.Ne5 Kf2 3.Nc4 Kf3 4.Kf6 (9.948)

8/14   0:00   +30.00   1.Nf7 Kg2 2.Kf6 Kf2 3.Ke5 Ke3 4.Be4 Kd2 (16.568)

8/16   0:00   +30.10   1.Kf6 Kg2 2.Ke5 Kf3 3.Kd4 Ke2 4.Nf7 Kd2 (18.110)

9/16   0:00   +30.10   1.Kf6 Kg2 2.Ke5 Kf3 3.Kd4 Kg4 4.Ke4 Kg5
                       5.Ke5 (28.091)

10/18  0:00   +30.20   1.Kf6 Kg2 2.Ke5 Kf2 3.Ke4 Ke2 4.Nf7 Kd2
                       5.Kd4 Kc1 (55.310)

11/18  0:00   +30.30   1.Kf6 Kg2 2.Ke5 Kf2 3.Ke4 Ke2 4.Nf7 Kd2
                       5.Kd4 Kc1 6.Ke3 (99.596)

12/20  0:00   +30.30   1.Kf6 Kg2 2.Nf7 Kf2 3.Ke5 Ke3 4.Nd6 Kd2
                       5.Kd4 Ke2 6.Nc4 Kf3 (182.816)

13/22  0:00   +30.30   1.Kf6 Kg2 2.Nf7 Kf2 3.Ke5 Ke3 4.Nd6 Kd2
                       5.Kd4 Ke2 6.Nc4 Kf3 7.Ke5 (325.890)

14/22  0:00   +30.30   1.Kf6 Kg2 2.Nf7 Kf2 3.Ke5 Ke3 4.Nd6 Kd2
                       5.Kd4 Ke2 6.Nc4 Kd1 7.Ke3 Kc1 (505.773)

15/24  0:01   +30.30   1.Kf6 Kg2 2.Nf7 Kf2 3.Ke5 Ke3 4.Nd6 Kd2
                       5.Kd4 Ke2 6.Nc4 Kd1 7.Ke3 Kc1 8.Be4 (951.239) 672

15/24  0:01   +30.50   1.Nf7 Kg2 2.Nd6 Kf3 3.Nc4 Kf4 4.Kf6 Kf3
                       5.Kg5 Ke2 6.Bc2 Ke1 7.Kf5 Ke2 8.Ke4 (1.248.929) 672

16/26  0:03   +30.50   1.Nf7 Kg1 2.Nd6 Kf2 3.Nc4 Kf3 4.Kf6 Kf4
                       5.Bf5 Kf3 6.Ke5 Ke2 7.Bc2 Kf2 8.Kf4 Ke2 (2.006.034) 667

17/28  0:04   +30.50   1.Nf7 Kg1 2.Nd6 Kf2 3.Nc4 Kf3 4.Kf6 Kf4
                       5.Bf5 Kf3 6.Ke5 Ke2 7.Bc2 Kf2 8.Kf4 Ke2
                       9.Ke4 (2.899.224) 666

17/28  0:05   +30.90   1.Kf6 Kg2 2.Ke5 Kf2 3.Kd4 Kf3 4.Be4+ Kg4
                       5.Nf7 Kf4 6.Kd3 Kg3 7.Ke3 Kg4 8.Bd3 Kh4
                       9.Be2 Kg3 (3.519.269) 665

18/28  0:06   +31.10   1.Kf6 Kg2 2.Ke5 Kf2 3.Kd4 Kf3 4.Be4+ Kg4
                       5.Nf7 Kf4 6.Kd3 Kg3 7.Ke3 Kg4 8.Bd3 Kh5
                       9.Kf4 Kh4 10.Kf3 (4.653.655) 669

19/32  0:10   +31.50   1.Kf6 Kg2 2.Ke5 Kf2 3.Kd4 Kf3 4.Be4+ Kg4
                       5.Nf7 Kf4 6.Kd3 Kg3 7.Ke3 Kg4 8.Bd3 Kh5
                       9.Kf4 Kh4 10.Be2 Kh3 (7.394.550) 678

20/32  0:15   +31.90   1.Kf6 Kg2 2.Ke5 Kf2 3.Kd4 Kf3 4.Be4+ Kg4
                       5.Nf7 Kf4 6.Kd3 Kg3 7.Ke3 Kg4 8.Bg6 Kg3
                       9.Bf5 Kg2 10.Bd3 Kh2 11.Kf4 (10.857.450) 689

21/36  0:23   +31.90   1.Kf6 Kg2 2.Ke5 Kf2 3.Kd4 Kf3 4.Be4+ Kg4
                       5.Nf7 Kf4 6.Kd3 Kg3 7.Ke3 Kg4 8.Bg6 Kg3
                       9.Ne5 Kh2 10.Nf3+ Kh1 11.Ne1 Kh2
                       12.Kf4 (16.296.714) 705

22/38  0:33   +31.90   1.Kf6 Kg2 2.Ke5 Kf2 3.Kd4 Kf3 4.Be4+ Kg4
                       5.Nf7 Kf4 6.Kd3 Kg3 7.Ke3 Kg4 8.Bg6 Kg3
                       9.Bf5 Kg2 10.Bd3 Kh2 11.Ke4 Kg1
                       12.Ke3 (23.841.117) 711

23/38  0:46   +32.30   1.Kf6 Kg2 2.Ke5 Kf2 3.Kd4 Kf3 4.Nf7 Ke2
                       5.Ne5 Kd2 6.Nc4+ Kc1 7.Bd3 Kd1 8.Bf5 Ke1
                       9.Bg4 Kf2 10.Ke4 Kg2 11.Ke3 Kh2
                       12.Kf2 Kh1 (33.296.339) 721

24/40  1:07   +32.30   1.Kf6 Kg2 2.Ke5 Kf2 3.Kd4 Kf3 4.Nf7 Ke2
                       5.Ne5 Kd2 6.Nc4+ Kc1 7.Bd3 Kd1 8.Bf5 Ke1
                       9.Bg4 Kf2 10.Ke4 Kg2 11.Ke3 Kh2
                       12.Kf2 Kh1 13.Kf3 (49.331.001) 727

25/40  1:27   +M15     1.Kf6 Kg2 2.Ke5 Kf2 3.Kd4 Kf3 4.Be4+ Kg4
                       5.Ke5 Kg3 6.Nf7 Kh2 7.Kf4 Kh3 8.Bd3 Kh4
                       9.Be2 Kh3 10.Ng5+ Kh2 11.Kf3 Kg1
                       12.Kg3 Kh1 13.Bd3 Kg1 14.Nh3+ (64.900.327) 737

26/40  1:51   +M15     1.Kf6 Kg2 2.Ke5 Kf2 3.Kd4 Kf3 4.Be4+ Kg4
                       5.Ke5 Kg3 6.Nf7 Kh2 7.Bd3 Kg3 8.Ke4 Kh4
                       9.Be2 Kg3 10.Ke3 Kg2 11.Ng5 Kh2
                       12.Kf2 Kh1 13.Kg3 Kg1 14.Nh3+ (84.086.192) 751

27/40  2:29   +M15     1.Kf6 Kg2 2.Ke5 Kf2 3.Kd4 Kf3 4.Be4+ Kg4
                       5.Ke5 Kg3 6.Nf7 Kh2 7.Bd3 Kg3 8.Ke4 Kh4
                       9.Be2 Kg3 10.Ke3 Kg2 11.Ng5 Kh2
                       12.Kf2 Kh1 13.Kg3 Kg1 14.Nh3+ (113.189.791) 756

27/40  2:43   +M14     1.Nf7 Kg1 2.Kf6 Kf2 3.Ke5 Ke3 4.Ng5 Kd2
                       5.Kd4 Ke2 6.Bh7 Kf2 7.Kd3 Ke1 8.Bf5 Kf1
                       9.Ne4 Kg1 10.Ke2 Kh2 11.Kf2 Kh1
                       12.Nd2 Kh2 13.Nf1+ Kh1 14.Be4+ (123.995.759) 757

28/40  3:09   +M14     1.Nf7 Kg1 2.Kf6 Kf2 3.Ke5 Ke3 4.Ng5 Kd2
                       5.Kd4 Ke2 6.Bh7 Kf2 7.Kd3 Ke1 8.Bf5 Kf1
                       9.Ne4 Kg1 10.Ke2 Kh2 11.Kf2 Kh1
                       12.Nd2 Kh2 13.Nf1+ Kh1 14.Be4+ (145.158.651) 765

29/40  3:53   +M14     1.Nf7 Kg1 2.Kf6 Kf2 3.Ke5 Ke3 4.Ng5 Kd2
                       5.Kd4 Ke2 6.Bh7 Kf2 7.Kd3 Ke1 8.Bf5 Kf1
                       9.Ne4 Kg1 10.Ke2 Kh2 11.Kf2 Kh1
                       12.Nd2 Kh2 13.Nf1+ Kh1 14.Be4+ (181.028.059) 773

29/40  4:15   +M13     1.Kf6 Kg2 2.Be4+ Kg3 3.Ke5 Kg4 4.Nf7 Kg3
                       5.Nd6 Kg4 6.Nf5 Kg5 7.Ne3 Kh4 8.Bd3 Kh3
                       9.Kf4 Kh4 10.Be2 Kh3 11.Bc4 Kh4
                       12.Nf5+ Kh5 13.Bf7+ (199.558.731) 781

30/40  4:45   +M13     1.Kf6 Kg2 2.Be4+ Kg3 3.Ke5 Kg4 4.Nf7 Kg3
                       5.Nd6 Kg4 6.Nf5 Kg5 7.Ne3 Kh4 8.Bd3 Kh3
                       9.Kf4 Kh4 10.Be2 Kh3 11.Bc4 Kh4
                       12.Nf5+ Kh5 13.Bf7+ (228.868.604) 800

best move: Kg7-f6 time: 4:46.563 min  n/s: 801.561  CPU 100.0%   n/s(1CPU): 801.561  nodes: 229.660.000

These mate values are nonsensical.
I am using ValueWin = 3000 as a base value for KBNK situations but maybe it got too close to ValueMate? (also because of some weird Futility pruning I'm doing maybe)

Eelco

With ValueWin = 10000 I now get a better result:


[Event "Search Depth 16 Plies"]
[Site "Engine Match"]
[Date "2008.03.05"]
[Round "1"]
[White "Toga 1.5 Checkov_beta1 bb5men"]
[Black "Shredder 11 Default"]
[Result "1-0"]
[SetUp "1"]
[FEN "7N/6K1/6B1/8/8/8/8/7k w - -"]

1. Nf7 {+30.50/16 1s} Kg1 {-M21/4 0s (Kg2)} 2. Nd6
{+30.20/16 2s (Ne5)} Kf1 {-M20/4 0s} 3. Kf6 {+30.90/16 2s
(Nc4)} Ke1 {-M21/4 0s} 4. Nc4 {+30.70/16 2s (Ke5)} Kd1
{-M21/4 0s} 5. Ke5 {+30.90/16 1s (Bd3)} Kc1 {-M20/4 0s}
6. Kd4 {+30.90/16 2s (Bd3)} Kd1 {-M19/4 0s} 7. Be4
{+31.00/16 4s (Bd3)} Kc1 {-M18/4 0s (Ke2)} 8. Kc3
{+31.20/16 1s} Kd1 {-M17/4 0s} 9. Ne3+ {+31.50/16 2s (Nd6)}
Kc1 {-M16/4 0s (Ke2)} 10. Nf5 {+31.80/16 0s} Kd1 {-M15/4
0s} 11. Nd4 {+31.90/16 1s (Ng3)} Kc1 {-M14/4 0s (Ke1)}
12. Ne2+ {+32.30/16 2s} Kd1 {-M13/4 0s} 13. Kd3 {+32.30/16
0s} Ke1 {-M12/4 0s} 14. Ke3 {+M12/16 0s} Kd1 {-M11/4 0s}
15. Bf5 {+M11/16 1s (Bb1)} Ke1 {-M10/4 0s} 16. Bc2 {+M10/16
1s} Kf1 {-M9/4 0s} 17. Nf4 {+M9/16 0s} Ke1 {-M8/4 0s}
18. Ng2+ {+M8/16 0s} Kf1 {-M7/4 0s} 19. Kf3 {+M7/16 0s} Kg1
{-M6/4 0s} 20. Kg3 {+M6/16 0s (Bf5)} Kf1 {-M5/4 0s}
21. Bd3+ {+M5/16 0s (Bd1)} Kg1 {-M4/4 0s} 22. Nf4 {+M4/16
0s (Be2)} Kh1 {-M3/4 0s} 23. Be2 {+M3/16 0s} Kg1 {-M2/4 0s}
24. Nh3+ {+M2/16 0s} Kh1 {-M1/4 0s} 25. Bf3# {+M1/16 0s} 1-0

Mate with Bishop and Knight against Lone King may not be of much practical value in tournaments but it sure is nice to see when a piece of programming  works out the way you'd like :) I could not have made it on my own, but the subroutine was already halfway finished by Thomas. The subroutine may not be very elegant and needs more testing but hopefully I learn something from it. 
Parent - - By Vempele (Silver) Date 2008-03-05 08:40

>About the bitbase probing, typical scores are or at least were around 40+. This exceeds ValueMate in Toga CMLX which I think could be a problem?


I just forgot to change it back after one of my ideas didn't work out. I'll try it again in 5e - it should be good for a few Elo. Adjusting bitbase scores (and material eval) is not a problem.
Parent - - By Eelco de Groot (***) Date 2009-07-05 18:37

>> About the bitbase probing, typical scores are or at least were around 40+. This exceeds ValueMate in Toga CMLX which I think could be a problem?
> I just forgot to change it back after one of my ideas didn't work out. I'll try it again in 5e - it should be good for a few Elo. Adjusting bitbase scores (and material eval) is not a problem.


We were vèèèry patient and now it is finally here :cool: Toga CMLX 1.4.5e! Tried your Microsoft build, seems okay! Good to see that the bitbases are also almost working, Vempele, thanks! There is maybe just a small issue with plus and minus I think?

8/8/2k5/8/8/8/4K3/5R2 b - -


Engine: Toga CMLX 1.4.5e bb (64 MB)
by Teemu Pudas, Thomas Gaksch and Fabien L

1/01   0:00   +M542--  1...Kb5 (3)

2/01   0:00   +M542--  1...Kb5 (20)

3/02   0:00   +M491--  1...Kb5 2.Kd1 (32)

4/02   0:00   +M491--  1...Kb5 2.Kd1 (58)

5/03   0:00   +M540--  1...Kb5 2.Kd1 Kc5 (136)

6/04   0:00   +M490--  1...Kb5 2.Kd1 Kc5 3.Rf5+ (409)

7/04   0:00   +M490--  1...Kb5 2.Kd1 Kc5 3.Rf5+ (713)

8/05   0:00   +M490--  1...Kb5 2.Kd1 Kc5 (1.136)

9/06   0:00   +M488--  1...Kb5 2.Kd1 Kc5 3.Rf5+ Kd6 4.Rf6+ (2.100)

10/06  0:00   +M488--  1...Kb5 2.Kd1 Kc5 3.Rf5+ Kd6 4.Rf6+ (3.403)

11/07  0:00   +M488--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 (5.295)

12/08  0:00   +M487--  1...Kb5 2.Kd1 Kc5 3.Rf5+ Kd6 4.Rf6+ Ke5
                       5.Rf7 (7.771)

13/08  0:00   +M486--  1...Kb5 2.Kd1 Kc5 3.Rf5+ Kd6 4.Rf6+ Kc5 (10.603)

14/09  0:00   +M485--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kc5 4.Rf5+ Kd6 (14.004)

15/10  0:00   +M486--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kd5 6.Rf6 (18.811)

16/10  0:00   +M486--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kc6
                       5.Rf7 Kd5 6.Rf6 (24.948)

17/11  0:00   +M486--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kd5 (32.248)

18/12  0:00   +M486--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kc6
                       5.Rf7 Kb6 6.Kc1 Kc5 7.Kd1 (42.058)

19/12  0:00   +M486--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kc6
                       5.Rf7 Kb6 6.Kc1 Kc5 7.Kd1 (53.809)

20/13  0:00   +M488--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kc5 (67.740)

21/14  0:00   +M484--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kc5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd6 7.Rf5 Ke6 8.Rf1 (79.650)

22/14  0:00   +M484--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kc5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd6 7.Rf5 Ke6 8.Rf1 (91.843)

23/15  0:00   +M488--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kc5 4.Rf5+ Kb6 (105.377)

24/16  0:00   +M483--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kc5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd6 7.Rf5 Ke6 8.Rf1 Ke7
                       9.Kd1 (122.068)

25/16  0:00   +M483--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Ke4
                       5.Rf7 Kd4 6.Kc1 Ke5 7.Rf1 Kd5 8.Rf5+ Kd6
                       9.Rf1 (144.325)

26/17  0:00   +M483--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Ke4
                       5.Rf7 Kd4 6.Kc1 Ke5 7.Rf1 Kd5 (174.802)

27/18  0:00   +M476--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd6 7.Rf5 Ke6 8.Rf1 Ke7 (204.024)

28/18  0:00   +M462--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd6 7.Rf5 Ke6 8.Rf1 Ke7 (237.428)

29/19  0:00   +M476--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd6 7.Rf5 Ke6 8.Rf1 Ke7 (274.113)

30/20  0:00   +M476--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd6 7.Rf5 Ke6 8.Rf1 Ke7 (317.995)

31/20  0:00   +M476--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd6 7.Rf5 Ke6 8.Rf1 Ke7 (363.716)

32/21  0:00   +M476--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd6 7.Rf5 Ke6 8.Rf1 Ke7 (413.282)

33/22  0:00   +M476--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd6 7.Rf5 Ke6 8.Rf1 Ke7 (467.717)

34/22  0:00   +M476--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd6 7.Rf5 Ke6 8.Rf1 Ke7 (527.874)

35/23  0:00   +M476--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd6 7.Rf5 Ke6 8.Rf1 Ke7 (602.825)

36/24  0:00   +M476--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd6 7.Rf5 Ke6 8.Rf1 Ke7 (675.558)

37/24  0:00   +M476--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kc5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd6 7.Rf5 Ke6 8.Rf1 Ke7 (747.183)

38/25  0:00   +M462--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd5 7.Rf5+ Ke6 8.Rf1 Ke7 (819.760)

39/26  0:00   +M451--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kc5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd5 7.Rf5+ Ke6 8.Rf1 Kd5
                       9.Rf6 Kc5 10.Rf4 Kd5 11.Kd1 Ke6
                       12.Rf8 Ke7 13.Rf1 Kd6 (881.863)

40/26  0:00   +M449--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd5 7.Rf5+ Ke6 8.Rf1 Ke7 (955.552)

41/27  0:00   +M455--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd5 7.Rf5+ Ke6 8.Rf1 Ke7 (1.041.621)

42/28  0:00   +M455--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd5 7.Rf5+ Ke6 8.Rf1 Ke7 (1.145.360)

43/28  0:00   +M455--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd5 7.Rf5+ Ke6 8.Rf1 Ke7 (1.253.264)

44/29  0:00   +M455--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kc5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd6 7.Rf5 Ke6 8.Rf1 Ke7 (1.366.952)

45/30  0:00   +M438--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd5 7.Rf5+ Ke6 8.Rf1 Ke7 (1.488.401)

46/30  0:00   +M443--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd5 7.Rf5+ Ke6 8.Rf1 Ke7 (1.620.859)

47/31  0:00   +M443--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd5 7.Rf5+ Ke6 8.Rf1 Ke7 (1.767.333)

48/32  0:00   +M415--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd5 7.Rf5+ Ke6 8.Rf1 Kd5
                       9.Rf6 Kc5 10.Rf4 Kd5 11.Kd1 Ke6
                       12.Rf1 Ke5 13.Re1+ Kf6 14.Kc1 Kg7 (1.939.467)

49/32  0:01   +M395--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd5 7.Rf5+ Ke6 8.Rf1 Kd5
                       9.Rf6 Kc5 10.Rf4 Kd5 11.Kd1 Ke6
                       12.Rf1 Ke5 13.Re1+ Kf6 14.Kc1 Kg7 (2.123.490) 1991

50/33  0:01   +M375--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd5 7.Rf5+ Ke6 8.Rf1 Kd5
                       9.Rf6 Kc5 10.Rf4 Kd5 11.Kd1 Ke6
                       12.Rf1 Ke5 13.Re1+ Kf6 14.Kc1 Kg7 (2.326.604) 1999

51/34  0:01   +M355--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd5 7.Rf5+ Ke6 8.Rf1 Kd5
                       9.Rf6 Kc5 10.Rf4 Kd5 11.Kd1 Ke6
                       12.Rf1 Ke5 13.Re1+ Kf6 14.Kc1 Kg7 (2.544.938) 2023

52/34  0:01   +M353--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd5 7.Rf5+ Ke6 8.Rf1 Ke7 (2.768.485) 2022

53/35  0:01   +M353--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd5 7.Rf5+ Ke6 8.Rf1 Ke7 (3.012.052) 2009

54/36  0:01   +M323--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd5 7.Rf5+ Ke6 8.Rf1 Kd5
                       9.Rf6 Kc5 10.Rf4 Kd5 11.Kd1 Ke6
                       12.Rf8 Ke7 13.Rf1 Kd8 14.Rf5 Ke8 (3.251.309) 2026

55/36  0:01   +M289--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd5 7.Rf5+ Ke6 8.Rf1 Kd5
                       9.Rf6 Kc5 10.Rf4 Kd5 11.Kd1 Ke6
                       12.Rf1 Ke5 13.Re1+ Kf6 14.Kc1 Kg7 (3.477.240) 2025

56/37  0:01   +M263--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd6 7.Rf5 Ke6 8.Rf1 Ke7 (3.722.396) 2035

57/38  0:01   +M237--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd5 7.Rf5+ Ke6 8.Rf1 Ke7 (3.980.906) 2037

58/38  0:02   +M237--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kd5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd5 7.Rf5+ Ke6 8.Rf1 Ke7 (4.253.301) 2028

59/39  0:02   +M237--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kc5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd5 7.Rf5+ Ke6 8.Rf1 Ke7 (4.544.262) 2043

60/40  0:02   +M172--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kc5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd5 7.Rf5+ Ke6 8.Rf1 Kd5
                       9.Rf6 Kc5 10.Rf4 Kd5 11.Kd1 Ke6
                       12.Rf1 Ke5 13.Re1+ Kf6 14.Kc1 Kg7 (4.848.547) 2045

61/40  0:02   +M130--  1...Kb5 2.Kd1 Kc6 3.Rf6+ Kc5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd6 7.Rf5 Ke6 8.Rf1 Kd5
                       9.Rf6 Kc5 10.Rf4 Kd5 11.Kd1 Ke6
                       12.Rf1 Ke5 13.Re1+ Kf6 14.Kc1 Kg7 (5.217.500) 2039

62/41  0:02   +M92--   1...Kb5 2.Kd1 Kc6 3.Rf6+ Kc5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd6 7.Rf5 Ke6 8.Rf1 Kd5
                       9.Rf6 Kc5 10.Rf4 Kd5 11.Kd1 Ke6
                       12.Rf8 Ke7 13.Rf1 Kd6 (5.592.366) 2044

63/42  0:02   +M46--   1...Kb5 2.Kd1 Kc6 3.Rf6+ Kc5 4.Rf5+ Kd6
                       5.Rf7 Kc5 6.Kc1 Kd5 7.Rf5+ Ke6 8.Rf1 Kd5
                       9.Rf6 Kc5 10.Rf4 Kd6 11.Rf8 Ke6
                       12.Re8+ Kd6 13.Kd1 Kc5 14.Rf8 Kd5 (5.921.697) 2042

best move: Kc6-b5 time: 4:00.000 min  n/s: 2.034.653  CPU 100.0%   n/s(1CPU): 2.034.653  nodes: 6.167.033

This is really supposed to be -M13 and 1...Kb5 -M11, 1...Kb5 is one of the three worst moves, so I think there is somewhere a +/- issue. But then again the bitbases are not necessary and will not improve rating, so it is not very important, at least the bitbases and dll get loaded now so good chances it can be fixed!

Thanks!

Eelco
Parent - By Vempele (Silver) Date 2009-07-05 18:43
Oops, looks like I forgot about everything I promised for 5e. Including changing ValueMate back.
Parent - - By Vempele (Silver) Date 2009-07-05 19:28
I've updated it now (twice - there was another pesky FEN bug after the first update).
Parent - - By Eelco de Groot (***) Date 2009-07-05 20:35

> I've updated it now (twice - there was another pesky FEN bug after the first update).


Thanks Vempele! Fast work and it is looking better now, same endgame position:

8/8/2k5/8/8/8/4K3/5R2 b - -


Engine: Toga CMLX 1.4.5e bb (64 MB)
by Teemu Pudas, Thomas Gaksch and Fabien L

1/01   0:00   -50.46   1...Kb5 (2)

1/01   0:00   -50.45   1...Kc5 (3)

1/01   0:00   -50.44   1...Kd5 (4)

2/01   0:00   -50.44   1...Kd5 (11)

3/02   0:00   -49.05   1...Kd5 2.Kd3 (43)

4/02   0:00   -49.05   1...Kd5 2.Kd3 (88)

5/03   0:00   -49.65   1...Kd5 2.Kd3 Ke5 (166)

6/04   0:00   -48.25   1...Kd5 2.Kd3 Ke5 3.Kc3 (425)

7/04   0:00   -48.25   1...Kd5 2.Kd3 Ke5 3.Kc3 (732)

8/05   0:00   -48.85   1...Kd5 2.Kd3 Ke5 3.Kc3 Ke6 (1.604)

9/06   0:00   -47.46   1...Kd5 2.Ke3 Ke5 3.Rd1 Ke6 4.Kd4 (5.233)

9/06   0:00   -47.45   1...Kd6 2.Kd3 Ke5 3.Kc3 Kd5 4.Ra1 (6.112)

10/06  0:00   -47.46   1...Kd6 2.Ke3 Ke5 3.Rd1 Ke6 4.Kd4 (9.160)

11/07  0:00   -48.06   1...Kd6 2.Ke3 Ke5 3.Rd1 Ke6 4.Kd4 Kd6 (16.170)

12/08  0:00   -46.66   1...Kd6 2.Ke3 Ke5 3.Rd1 Kf5 4.Kd4 Ke6
                       5.Kc4 (35.202)

13/08  0:00   -46.66   1...Kd6 2.Ke3 Ke5 3.Rd1 Kf5 4.Kd4 Ke6
                       5.Kc4 (55.200)

14/09  0:00   -47.26   1...Kd6 2.Ke3 Ke5 3.Rd1 Kf5 4.Kd4 Ke6
                       5.Ke4 Ke7 (85.596)

15/10  0:00   -45.87   1...Kd6 2.Ke3 Ke5 3.Rd1 Kf5 4.Kd4 Ke6
                       5.Ke4 Ke7 6.Kd5 (160.683)

15/10  0:00   -45.86   1...Kd5 2.Rf5+ Ke4 3.Ra5 Kd4 4.Kf3 Kc4
                       5.Ke4 Kb4 6.Rf5 (178.683)

16/10  0:00   -45.86   1...Kd5 2.Rf5+ Ke4 3.Ra5 Kd4 4.Kf3 Kc4
                       5.Ke4 Kb4 6.Rf5 (238.947)

17/11  0:00   -46.46   1...Kd5 2.Rf5+ Ke4 3.Ra5 Kd4 4.Kf3 Kc4
                       5.Ke4 Kb4 6.Rf5 Kc4 (326.682)

18/12  0:01   -45.07   1...Kd5 2.Kd3 Ke5 3.Ke3 Kd5 4.Rf5+ Kd6
                       5.Ke4 Ke6 6.Rd5 Ke7 7.Ke5 (519.907) 420

18/12  0:01   -45.06   1...Kd6 2.Rf5 Ke6 3.Ra5 Kd6 4.Kf3 Ke7
                       5.Kf4 Ke6 6.Rc5 Kd6 7.Re5 (577.415) 420

19/12  0:01   -45.07   1...Kd6 2.Rf5 Ke6 3.Ra5 Kd6 4.Kf3 Ke7
                       5.Ke4 Ke6 6.Rd5 Ke7 7.Ke5 (732.032) 424

20/13  0:02   -45.67   1...Kd6 2.Rf5 Ke6 3.Ra5 Kd6 4.Kf3 Ke7
                       5.Ke4 Ke6 6.Ra6+ Ke7 7.Kd5 Kd7 (969.995) 429

21/14  0:03   -44.28   1...Kd6 2.Kd3 Ke5 3.Re1+ Kd5 4.Kc3 Kc6
                       5.Rd1 Kc5 6.Rd4 Kc6 7.Kc4 Kc7 8.Kd5 (1.410.372) 432

22/14  0:04   -44.27   1...Kd6 2.Kd3 Ke5 3.Re1+ Kd5 4.Kc3 Kc6
                       5.Rd1 Kc5 6.Rd4 Kb6 7.Rc4 Kb5 8.Kd4 (1.846.269) 435

23/15  0:05   -44.87   1...Kd6 2.Kd3 Ke5 3.Re1+ Kd5 4.Kc3 Kc6
                       5.Rd1 Kc5 6.Rd4 Kb6 7.Rc4 Kb5 8.Kd4 Ka5 (2.310.697) 437

24/16  0:08   -43.48   1...Kd6 2.Kd3 Ke5 3.Re1+ Kd5 4.Kc3 Kc6
                       5.Rd1 Kc5 6.Rd4 Kb6 7.Rc4 Kb5 8.Kd4 Ka5
                       9.Kc3 (3.693.668) 440

24/16  0:08   -43.46   1...Kd5 2.Rf5+ Ke6 3.Rh5 Ke7 4.Ke3 Ke6
                       5.Ke4 Kd6 6.Re5 Kc6 7.Rd5 Kb6 8.Kf3 Kc6
                       9.Rf5 (3.894.281) 440

25/16  0:10   -43.48   1...Kd5 2.Kd3 Ke5 3.Re1+ Kd5 4.Kc3 Kc6
                       5.Rd1 Kc5 6.Rd4 Kb6 7.Rc4 Kb5 8.Kd4 Ka5
                       9.Kc3 (4.761.271) 441

26/17  0:13   -44.08   1...Kd5 2.Kd3 Ke5 3.Re1+ Kd5 4.Kc3 Kc6
                       5.Rd1 Kc5 6.Rd4 Kb6 7.Rc4 Kb5 8.Kd4 Kb6
                       9.Rc5 Ka6 (5.790.543) 442

27/18  0:25   -42.69   1...Kd5 2.Kd3 Ke5 3.Re1+ Kd5 4.Kc3 Kc6
                       5.Rd1 Kc5 6.Rd4 Kb6 7.Rc4 Kb5 8.Kd4 Kb6
                       9.Rc5 Ka6 10.Kc4 (11.004.987) 437

28/18  0:29   -42.69   1...Kd5 2.Kd3 Ke5 3.Re1+ Kd5 4.Kc3 Kc6
                       5.Kd4 Kd6 6.Re5 Kc7 7.Kc5 Kd7 8.Re1 Kd8
                       9.Kd6 Kc8 10.Re5 (13.072.185) 439

28/18  0:31   -42.68   1...Kd6 2.Rd1+ Ke6 3.Ke3 Kf5 4.Rd5+ Kf6
                       5.Kf4 Ke6 6.Ke4 Kf6 7.Re5 Kg6 8.Rf5 Kh6
                       9.Ke5 Kg6 10.Kf4 (13.825.401) 439

29/19  0:36   -43.28   1...Kd6 2.Rd1+ Ke6 3.Ke3 Kf5 4.Rd5+ Kf6
                       5.Kf4 Ke6 6.Ke4 Kf6 7.Re5 Kg6 8.Rf5 Kh6
                       9.Ke5 Kg6 10.Kf4 Kg7 (16.053.380) 440

30/20  0:56   -41.90   1...Kd6 2.Kd3 Ke5 3.Re1+ Kd5 4.Re4 Kd6
                       5.Kd4 Kc7 6.Kd5 Kb6 7.Ra4 Kb5 8.Rc4 Kb6
                       9.Rc5 Ka6 10.Kc6 Ka7 11.Re5 (24.145.299) 428

31/20  1:10   -41.89   1...Kd6 2.Kd3 Ke5 3.Re1+ Kd5 4.Re4 Kd6
                       5.Kd4 Kc7 6.Kd5 Kb6 7.Kd6 Kb5 8.Rd4 Kb6
                       9.Rd5 Ka7 10.Kc6 Ka6 11.Rh5 (29.747.158) 422

31/20  1:12   -41.88   1...Kd5 2.Ke3 Ke5 3.Re1 Kd5 4.Kd3 Kc5
                       5.Ke4 Kd6 6.Kd4 Kc6 7.Re6+ Kb5 8.Kd5 Kb4
                       9.Rc6 Kb5 10.Rc4 Kb6 11.Rc5 (30.910.995) 423

32/21  1:42   -42.49   1...Kd5 2.Kd3 Ke5 3.Re1+ Kf6 4.Re4 Kg5
                       5.Ke3 Kf5 6.Kf3 Kg5 7.Rf4 Kg6 8.Kg4 Kh6
                       9.Rf6+ Kh7 10.Kf5 Kg7 11.Kg5 Kg8 (43.893.907) 429

33/22  3:04   -41.11   1...Kd5 2.Kd3 Ke5 3.Re1+ Kf6 4.Re4 Kg5
                       5.Ke3 Kf5 6.Kf3 Kg5 7.Rf4 Kg6 8.Kg4 Kh6
                       9.Kf5 Kg7 10.Kg5 Kg8 11.Kg6 Kh8
                       12.Rf7 (79.445.595) 429

34/22  7:50   -41.11   1...Kd5 2.Kf3 Kd4 3.Kf4 Kc4 4.Ke4 Kc3
                       5.Rf7 Kc4 6.Rc7+ Kb5 7.Kd4 Kb4 8.Rc5 Kb3
                       9.Rc4 Ka3 10.Kc3 Ka2 11.Rc7 Ka1
                       12.Rc5 (200.459.524) 426

35/23  34:33  -41.71   1...Kd5 2.Kf3 Kd4 3.Kf4 Kc4 4.Ke4 Kc3
                       5.Rf2 Kc4 6.Rc2+ Kb5 7.Kd4 Kb4
                       8.Rb2+ Ka3 9.Rb1 Ka2 10.Rb4 Ka3
                       11.Kc3 Ka2 12.Rb3 Ka1 (909.530.970) 438

best move: Kc6-d5 time: 35:29.922 min  n/s: 438.677  CPU 100.0%   n/s(1CPU): 438.677  nodes: 934.326.868


1... Kd5 and 1... Kd6 (or Kd7, Kc5) are correct moves because they postpone the mate the longest to - M13. Toga with bitbases has trouble finding the actual mate as you can see, but that is a known issue using Scorpio bitbases with Toga. All Bill Mowery's versions only use the probes that are draws and ignore the wins and losses in the bitbases scores. Often these do not seem to fit very well with Toga's eval not such a surprise because they are designed by Daniel Shawul for Scorpio and Toga does no effort to make any corrections to the probescores. It is not ideal but still useful for analysis purposes. I tried to do some things with the matescores in the probes to get better results, but have not tested it very much.

Regards, Eelco
Parent - By Vempele (Silver) Date 2009-07-06 11:38 Edited 2009-07-06 11:51
I've changed my mind on this issue. It's not worth fixing - online tablebases will cover everything once the root position has 6 men.
Parent - - By gerold (**) Date 2008-03-05 14:03
You are off my good guy list.
Parent - By Vempele (Silver) Date 2009-07-05 19:05
There. 1.4.5e supports Arena again. Am I back on your good guy list?
Parent - By FWCC (***) Date 2008-03-07 21:58
I'm still having problems with Toga simply just stopping in my Fritz 10 GUI and I have Vista.Adjusting Task Manager helps but not 100%.Can anyone help with an opinion?
Up Topic The Rybka Lounge / Computer Chess / Toga CMLX 1.4 beta 5c!

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill