Not logged inRybka Chess Community Forum
Up Topic The Rybka Lounge / Computer Chess / Komodo crushes Houdini in 25 moves from standard opening
- - By Josef (**) Date 2012-05-07 20:34
Hi,

what do you think about this miniature? Where does Houdini fail?

[Event "Eagle, Blitz:25'+10""]
[Site "Eagle"]
[Date "2012.05.05"]
[Round "15"]
[White "Komodo32 Version 4-c2"]
[Black "Houdini 2.0c w32"]
[Result "1-0"]
[ECO "B83"]
[Annotator "0.19;0.08"]
[PlyCount "49"]

{W=18.9 ply; 356kN/s  B=18.9 ply; 1.272kN/s} 1. e4 c5 2. Nf3 d6 3. d4 cxd4 4.
Nxd4 Nf6 5. Nc3 e6 {Both last book move} 6. Be2 {[%eval 19,18] [%emt 0:02:09]}
Be7 {(Nc6) [%eval 8,20] [%emt 0:00:43]} 7. O-O {
(Qd3) [%eval 1,0] [%emt 0:00:00]} O-O {[%eval 16,21] [%emt 0:00:54]} 8. f4 {
(Qd3) [%eval 1,0] [%emt 0:00:00]} e5 {[%eval 0,20] [%emt 0:00:45]} 9. Nf5 {
(Nb3) [%eval 14,19] [%emt 0:01:20]} Bxf5 {[%eval -10,19] [%emt 0:00:52]} 10.
exf5 {[%eval 13,21] [%emt 0:01:24]} Nc6 {[%eval -4,20] [%emt 0:02:33]} 11. Kh1
{(Nd5) [%eval 9,22] [%emt 0:01:52]} Rc8 {(h6) [%eval -12,17] [%emt 0:00:49]}
12. g4 {(Bf3) [%eval 28,18] [%emt 0:00:27]} Nd7 {
(exf4) [%eval 13,18] [%emt 0:00:57]} 13. Rg1 {(g5) [%eval 25,17] [%emt 0:01:27]
} Nd4 {[%eval -16,18] [%emt 0:00:53]} 14. Bd3 {[%eval 25,18] [%emt 0:00:18]}
Nc5 {(exf4) [%eval 2,19] [%emt 0:02:55]} 15. Bf1 {
(Be3) [%eval 39,16] [%emt 0:00:35]} h6 {[%eval -2,17] [%emt 0:00:41]} 16. Bg2 {
(Be3) [%eval 47,17] [%emt 0:02:00]} b5 {(Bh4) [%eval 0,18] [%emt 0:00:28]} 17.
Nd5 {(fxe5) [%eval 51,15] [%emt 0:00:24]} Nc6 {[%eval 22,18] [%emt 0:00:59]}
18. g5 {(Rf1) [%eval 110,16] [%emt 0:00:11]} hxg5 {
[%eval 187,16] [%emt 0:00:36]} 19. fxg5 {[%eval 208,19] [%emt 0:00:24]} Qd7 {
(Bxg5) [%eval 254,17] [%emt 0:01:01]} 20. f6 {
(b4) [%eval 220,18] [%emt 0:00:34]} Bd8 {[%eval 264,19] [%emt 0:00:42]} 21. Be3
{(fxg7) [%eval 247,19] [%emt 0:00:36]} Qf5 {[%eval 355,15] [%emt 0:00:46]} 22.
Rf1 {[%eval 319,18] [%emt 0:00:19]} Qh7 {[%eval 684,18] [%emt 0:01:45]} 23.
fxg7 {[%eval 375,19] [%emt 0:00:22]} Re8 {(Kxg7) [%eval 838,17] [%emt 0:01:26]}
24. Qg4 {(Bxc5) [%eval 514,17] [%emt 0:00:23]} Re6 {
(Rb8) [%eval 1032,16] [%emt 0:01:55]} 25. Bxc5 {[%eval 716,17] [%emt 0:00:36]}
1-0
Parent - - By Barnard (Bronze) Date 2012-05-07 21:11
you have the answer in your own post/analisis:

15...h6 is a mistake(weakening the castling,and giving a target to atack),is better to play for example 15...Re8

17...Nc6??,a horribly blunder,allowing white to play that line 18.g5,hxg5 19.fxg5,and whatever blacks move will be,white has a winning advantage...from that point,the game is won for white

so the lossing move is 17...Nc6??

i hope with that,i answered your question
Parent - - By Josef (**) Date 2012-05-07 21:53
Hi Jerry,

thanks for the answer. These 2 moves really look bad. I didn't expect that from Houdini and thought Komodo's attack was more forcing. But at move 15 before h6 Komodo only has a very slight edge as I understand now.

Seems there is still a lot to be done in the Houdini engine.

I also looked the opening up in the book and see that 8. ... e5 is not listed, so it might be weak (but Komodo would also play it).

After 9.Nf5 both engines consider it weak to allow 10.Nxe7.
rnbq1rk1/pp2bppp/3p1n2/4pN2/4PP2/2N5/PPP1B1PP/R1BQ1RK1 b - - 0 9

I'm not sure why since 9. ... Bxf5 also looses the Bishop pair for Black. Somehow the black square Bishop seems more important for defense.

Then, 10. ... exf would give Black an isolated d-pawn, so he cannot take f4.
rn1q1rk1/pp2bppp/3p1n2/4pP2/5P2/2N5/PPP1B1PP/R1BQ1RK1 b - - 0 10

White's pawns at f4, f5, g4 look dangerous to me, so I thought the attack was forcing, but Komodo only won through Houdini's blunders.

Best regards,
Josef
Parent - - By Barnard (Bronze) Date 2012-05-07 22:06
agree,Houdini played the opening weakly

something that i cant understand is why both engines were out of book at 5 move

what opening book they were using to play the game?
Parent - - By Josef (**) Date 2012-05-07 22:43
Hi Jerry,

I play with a set of opening positions, and since Komodo is (still) the weaker engine I additionally allow it its private opening book. 7.0-0 and 8.f4 were played from this book, 6.Be2 for unknown reason not, anyhow it's also a main book move. I have every engine play on one core of my (actually my company's which I use during weekends) 3 Ghz Core 2 Duo with pondering on. And, since there was some discussion, I definitely see Komodo getting closer to Houdini on longer time controls. Currently after 143 games with 25'+10'' and after sorting out games where Komodo stopped playing and lost on time (a major bug in this engine, but does not hurt the other games) the standing is +35/-52/=56 meaning 44% for Komodo, with relatively few draws. To my observation, Komodo's wins are more positional (like this one) and Houdini's more tactical, making it interesting to watch. I'm looking forward to multicore Komodo.

Best Regards,
Josef
Parent - - By Barnard (Bronze) Date 2012-05-07 22:54
interesting results

but if i understood you fine,Houdini is playing without an opening book (just with opening positions) and Komodo is playing with a opening book?im right or worng?
Parent - - By Pia (****) Date 2012-05-08 16:42
Both engines were using book moves from GUI up to 5 move, afterwards, Komodo used internal book up to 8 move.
Parent - By Barnard (Bronze) Date 2012-05-08 17:04
Right

And if you add that to the weak computer,you will have blunders like "h6" and "Nc6"
Parent - - By ernest (****) Date 2012-05-08 18:43

> so the lossing move is 17...Nc6??


Indeed, takes a long time until Houdini sees the light:
(but is 17...b4 a saving move?)

Komodo32 Version 4-c2 - Houdini 2.0c w32, Eagle, Blitz:25'+10 2012
2rq1rk1/p3bpp1/3p3p/1pnNpP2/3n1PP1/8/PPP3BP/R1BQ2RK b - - 0 1

Analysis by Houdini 2.0c w32 Nlm:
17...Nc6 18.b4 Nd7 19.a4 Nb6 20.axb5 Nd4 21.Ra2 Nxb5 22.Bb2 Bf6 23.Nxf6+ Qxf6 24.Ra6 Nc4 25.fxe5 Nxb2 26.exf6 Nxd1 27.Rxd1 Rxc2 28.fxg7 Kxg7
  =  (0.17)   Depth: 17/37   00:00:07  17286kN
17...Nc6 18.b4 Nd7 19.a4 bxa4 20.b5 Nd4 21.Rxa4 Nxb5 22.c4 Nd4 23.Rxa7 Bh4 24.Ba3 Nc5 25.Qd2 Re8 26.Rd1 Ncb3 27.Qc3 Nc5 28.Bxc5 Rxc5
  =  (0.23)   Depth: 18/41   00:00:11  28560kN
17...Nc6 18.b4
  +/=  (0.29 --)   Depth: 19/53   00:00:30  76714kN
17...Nc6 18.b4
  +/=  (0.42 --)   Depth: 19/53   00:00:50  130mN
17...Nc6 18.b4
  +/-  (0.76 --)   Depth: 19/64   00:01:17  198mN
17...b4 18.Nxb4 a5 19.Nd5 Re8 20.fxe5 dxe5 21.c3 Nb5 22.f6 Bxf6 23.Qf1 Nd6 24.Nxf6+ Qxf6 25.Qxf6 gxf6 26.Bxh6 e4 27.Rgf1 Re6 28.Bf4 Nc4 29.b3 Ne5 30.Bxe5 fxe5 31.Kg1 Rg6
  +/=  (0.56)   Depth: 19/64   00:01:38  256mN
17...b4 18.Nxb4
  +/=  (0.63 --)   Depth: 20/64   00:02:00  312mN
Parent - By Barnard (Bronze) Date 2012-05-08 18:49
Is more than + 0,63 , is near to +1,80 or even more the trully score

and for the move 17...b4,i dont know if it can save the game,i cant analize the move now,when i can i will take a look
Parent - - By Stonehenge (***) Date 2012-05-08 19:52
17...Nc6 certainly is very bad, but the position appears to be already lost for Black.
Below a long analysis with Houdini, using 3 cores of a Core i5-750 with 2048 MB of hash.

2rq1rk1/p3bpp1/3p3p/1pnNpP2/3n1PP1/8/PPP3BP/R1BQ2RK b - -


Engine: Houdini DEV 3T Scorpio (2048 MB)
by Robert Houdart

17/41  0:02   +0.23    1...Nc6 2.Qf1 Re8 3.b4 Nd7 4.c3 exf4
                       5.Bxf4 Nce5 6.a4 bxa4 7.Rxa4 Bg5
                       8.Rxa7 Bxf4 9.Qxf4 Rc4 10.Qd2 Qb8
                       11.Rga1 Rxg4 12.Ne3 Rg5 (13.525.741) 5267

18/42  0:04   +0.30--  1...Nc6 2.Qf1 (24.911.365) 5388

18/42  0:06   +0.43--  1...Nc6 2.Qf1 (35.108.427) 5433

18/52  0:15   +0.77--  1...Nc6 2.Qf1 (84.955.131) 5499

18/67  0:41   +1.99--  1...Nc6 2.Qf1 (229.811.675) 5485

18/67  0:55   +1.00    1...Nd7 2.c3 Nc6 3.Nxe7+ Qxe7 4.g5 hxg5
                       5.fxg5 e4 6.Bf4 Nc5 7.Bxd6 Nd3
                       8.Qxd3 exd3 9.Bxe7 Nxe7 10.Be4 Rc5
                       11.f6 Nd5 12.Bxd3 b4 13.cxb4 Nxb4
                       14.Bf1 g6 (301.104.240) 5470

18/67  1:02   +0.86    1...a5 2.fxe5 dxe5 3.c3 Nc6 4.Be3 Nb8
                       5.Qe2 Nbd7 6.Qxb5 Rb8 7.Qe2 Bg5
                       8.Bf2 Bh4 9.Bxh4 Qxh4 10.b4 Na4
                       11.Rac1 Rfd8 12.Rgd1 axb4 13.cxb4 Nab6
                       14.Nxb6 Nxb6 (339.584.441) 5470

19/67  1:07   +0.86    1...a5 2.fxe5 dxe5 3.c3 Nc6 4.Be3 Nb8
                       5.Qe2 Nbd7 6.Qxb5 Rb8 7.Qe2 Bg5
                       8.Bf2 Bh4 9.Bxh4 Qxh4 10.b4 Na4
                       11.Rac1 Rfd8 12.Rgd1 axb4 13.cxb4 Nab6
                       14.Nxb6 Nxb6 (367.697.587) 5474

20/67  1:18   +0.77++  1...a5 (433.604.010) 5502

20/67  1:21   +0.92    1...a5 2.fxe5 dxe5 3.c3 Nc6 4.Be3 Nd7
                       5.Qe2 Bg5 6.Qxb5 Bxe3 7.Nxe3 e4
                       8.Bxe4 Nce5 9.Rad1 Re8 10.Qe2 Qh4
                       11.Bd5 Nf6 12.Rd4 Qg5 13.Qf2 Nc6
                       14.Bxc6 Rxe3 (447.972.011) 5505

21/67  1:31   +0.92    1...a5 2.fxe5 dxe5 3.c3 Nc6 4.Be3 Nd7
                       5.Qe2 Bg5 6.Qxb5 Bxe3 7.Nxe3 e4
                       8.Bxe4 Nce5 9.Rad1 Re8 10.Qe2 Qh4
                       11.Bd5 Nf6 12.Rd4 Qg5 13.Qf2 Nc6
                       14.Bxc6 Rxe3 (507.602.305) 5520

22/67  1:57   +1.01--  1...a5 2.fxe5 (653.885.515) 5551

22/67  3:02   +1.19--  1...a5 2.fxe5 (1.018.527.359) 5588

22/67  3:23   +1.15    1...Nd7 2.c3 Nc6 3.Nxe7+ Qxe7 4.g5 hxg5
                       5.fxg5 e4 6.Bf4 Nce5 7.f6 Qe6 8.fxg7 Kxg7
                       9.Bxe4 Nc5 10.Bd5 Qf5 11.Be3 Ned3
                       12.Qf3 Qxf3+ 13.Bxf3 Rfe8 14.Bd4+ Ne5 (1.137.111.124) 5592

23/67  4:15   +1.15    1...Nd7 2.c3 Nc6 3.Nxe7+ Qxe7 4.g5 hxg5
                       5.fxg5 e4 6.Bf4 Nce5 7.f6 Qe6 8.fxg7 Kxg7
                       9.Bxe4 Nc5 10.Bd5 Qf5 11.Be3 Ned3
                       12.Qf3 Qxf3+ 13.Bxf3 Rfe8 14.Bd4+ Ne5 (1.430.402.927) 5601

24/67  5:42   +1.24--  1...Nd7 2.c3 (1.931.753.756) 5641

24/72  9:03   +1.42--  1...Nd7 2.c3 (3.078.333.011) 5667

24/72  14:01  +1.28    1...Nd7 2.Nxe7+ Qxe7 3.c3 Nc6 4.g5 hxg5
                       5.fxg5 e4 6.f6 Qe5 7.Rf1 e3 8.Qf3 e2
                       9.Re1 Rfe8 10.Bf4 Qf5 11.Bxd6 Qxf3
                       12.Bxf3 gxf6 13.Rxe2 Rxe2 14.Bxe2 fxg5 (4.784.017.861) 5685

25/72  14:47  +1.37--  1...Nd7 2.Nxe7+ (5.045.179.289) 5685

25/72  17:59  +1.58--  1...Nd7 2.Nxe7+ (6.166.090.569) 5709

25/95  26:24  +1.63    1...Nd7 2.Nxe7+ Qxe7 3.c3 Nc6 4.g5 hxg5
                       5.fxg5 Qe8 6.f6 g6 7.Qxd6 Nb6 8.b3 Qd7
                       9.Qxd7 Nxd7 10.Rd1 Ncb8 11.Ba3 Rfe8
                       12.Bh3 Rc7 13.Bf1 a6 14.Rac1 Nb6 (9.078.243.631) 5729

26/95  32:19  +1.63    1...Nd7 2.Nxe7+ Qxe7 3.c3 Nc6 4.g5 hxg5
                       5.fxg5 Qe8 6.f6 g6 7.Qxd6 Nb6 8.b3 Qd7
                       9.Qxd7 Nxd7 10.Rd1 Ncb8 11.Ba3 Rfe8
                       12.Bh3 Rc7 13.Bf1 a6 14.Rac1 Nb6 (11.160.105.843) 5753

Best move is: 17...Black resigns :eek:.
Parent - By Barnard (Bronze) Date 2012-05-08 20:00
Hi Robert

Thanks for the analisis

Can we agree that all the Houdini troubles begin when he played "h6",and all of that is due to the hardware used?
Parent - - By Nelson Hernandez (Gold) Date 2012-05-07 22:39
8...e5 wasn't a bad move, but it certainly was a strange move.  The kind of move you see two times in a thousand games from that position.  I have it with 40.5% success while 8...a6 is 48.3% and 8...Nc6 is 47.1%.
Parent - - By Barnard (Bronze) Date 2012-05-07 22:52
thanks Nelson

i have that move listed as interesting,but with only very few games in the book,and the better moves 8...a6 and 8...Nc6,the same than you...just i cant understand why Houdini played that move,if obviously 8...a6 and 8...Nc6 are better
Parent - - By Nelson Hernandez (Gold) Date 2012-05-07 22:59
It is not clear that they are better purely on merit.  What is clear is that they are both much more tested and for that reason alone are safer choices.  For all we know all three moves are forced draws.
Parent - By Barnard (Bronze) Date 2012-05-07 23:04
right

by better i mean better because are more tested,and that move is only played a very few times,that is why i told i have marked that move in my book as 'interesting' (meaning that it deserves to be investigated/played more times)

the line seems drawish in all the 3 cases,i must agree with that,and is the type of lines that i wont play (i dont like drawish lines,that one player only can take advantage if his opponent blunders,not by the investigation/innovation of the player)
Parent - - By Josef (**) Date 2012-05-08 22:59
Thanks, I didn't see this thread before. It shows that Komodo really overtakes Houdini with a certain amount of calculation time. I wondered if Komodo's getting closer is just an increase of draws, when it would never reach Houdini. That's not true, something is different in the Komodo engine which exploits calculation time more efficiently. No one can deny this fact.

Maybe the point is that at some depth Komodo stops getting outplayed tactically due to its slower search, when it wins positionally.
Parent - By NATIONAL12 (Gold) Date 2012-05-08 23:26
I did a similar test on Houdini 1.5a-Rybka 4 8 cpu at 40/40,this was all Houdini could do at the time.4.2 Ghz on different computer.
Result was +31 ELO to houdini,its further back on same computer chess thread.
Parent - - By M ANSARI (*****) Date 2012-05-08 06:07 Edited 2012-05-08 06:10
I think a lot has to do with the weak hardware being used.  White has a very imposing attack coming and the pawn rush on the King looks very dangerous, black has an OK setup but cannot play any passive moves.  The only problem with black is that he has his bishop and Queen running into each other.  While in many cases a bishop and Queen can be a real battering ram, here they seem not coordinated.

2rq1rk1/pp2bpp1/3p3p/2n1pP2/3n1PP1/2N5/PPP3BP/R1BQ2RK b - - 0 16


I think even in this late position black can make a good game out of it by playing the Bh4 with the idea of then Bf2 and Qh4 (probably Bh4 played earlier would have been better), this moves the white rook from the King g file removing the possibility of a future deadly discovered check and then black has the simple mate threat of Bg3 which makes white play h3 and weakens King position, and makes a good dynamic counter attack that is much better than passive play.  Most likely black would have to sac an exchange or a pawn but then white attack fizzles and he is left with a very exposed King that might be vulnerable and things can get quickly reveresed.  With a single core and 32 bits, such opening are bound to quickly offer a smother mate for one side as the horizon effect simply doesn't allow the engine to see impending danger that a human can quickly see.
Parent - - By BigBen (****) Date 2012-05-08 08:27
Hi,
   I cannot get my machine to reproduce moves such as 15...h6 [17ply] (Bh4) ...So the hardware is def playing a big part as in about 2 seconds my machine is hitting 20+ ply
Parent - By Barnard (Bronze) Date 2012-05-08 08:54
i can reproduce that moves,my machine is worst than him :smile: running at about 650-700 kn/s :smile:

lucky of me to have that dinosaur :lol:
Parent - - By Peter Grayson (****) Date 2012-05-08 17:01
I agree, my machine does not even consider 15..h6 but it follows the game right up to that point. In fact after 16.Bg2 its evaluation starts to favour white. Wonder if the engine was optimised.

Interestingly the w32 Pro version would play a different game when it did not consider 6..Be7 at all.

Wondering if something is wrong with the set up because with 1 thread, H2.0 w32 is giving similar kN/s but Komodo is showing about 20% faster.

PeterG

Komodo32 Version 4-c2 - Houdini 2.0c w32, 5'/40+5'/40+5'/40
rnbqkb1r/pp3ppp/3ppn2/8/3NP3/2N5/PPP1BPPP/R1BQK2R b KQkq - 0 1


Analysis by Houdini 2.0c Pro w32 GTB:

6...d5 7.exd5 Nxd5 8.Bb5+ Bd7 9.Nxd5 Bxb5 10.Nxb5 exd5 11.Qe2+ Be7 12.Bf4 Qa5+ 13.c3
  +/-  (0.84)   Depth: 6/20   00:00:00  10kN
6...Nc6 7.0-0 e5 8.Nb3 a6 9.Be3 Be6
  +/=  (0.50)   Depth: 6/21   00:00:00  15kN
6...Nc6 7.0-0 e5 8.Nb3 a6 9.Be3 Be6
  +/=  (0.50)   Depth: 7/21   00:00:00  18kN
6...e5
  +/=  (0.43 !)   Depth: 7/21   00:00:00  21kN
6...e5 7.Nf3 Nc6 8.0-0 h6 9.Be3 Be6 10.h3
  +/=  (0.48)   Depth: 7/25   00:00:00  28kN
6...e5 7.Nf3 Nc6 8.0-0 h6 9.Be3 Be6 10.h3 d5
  +/=  (0.42)   Depth: 8/25   00:00:00  31kN
6...e5 7.Nf3 Nc6 8.0-0 h6 9.Be3 Be6 10.h3 d5
  +/=  (0.42)   Depth: 9/25   00:00:00  40kN
6...e5 7.Nf5 Nc6 8.0-0 Bxf5 9.exf5 Qd7 10.Nd5 Nxd5 11.Qxd5 Qxf5 12.c3 Be7 13.Be3
  +/=  (0.36)   Depth: 10/28   00:00:00  85kN
6...e5
  +/=  (0.29 !)   Depth: 11/28   00:00:00  152kN
6...e5 7.Nf5 Nc6 8.0-0 Bxf5 9.exf5 Be7 10.Qd3 0-0 11.Bf3 Nb4 12.Qe2 Qd7 13.a3
  +/=  (0.32)   Depth: 11/28   00:00:00  165kN
6...e5
  =  (0.25 !)   Depth: 12/30   00:00:00  374kN
6...e5 7.Nf5 Nc6 8.0-0 Bxf5 9.exf5 Be7 10.Qd3 Nd4 11.Be3 Nxe2+ 12.Qxe2 0-0 13.Rad1 Qd7 14.Qd3
  =  (0.25)   Depth: 12/30   00:00:00  415kN
6...e5 7.Nf3 Nc6 8.0-0 Be7 9.Bg5 h6 10.Bxf6 Bxf6 11.Qd2 0-0 12.Rad1 Qb6 13.Nd5 Qxb2 14.Nxf6+ gxf6 15.Qxh6 Qxc2
  +/=  (0.26)   Depth: 13/31   00:00:00  922kN
6...e5 7.Nf3 Nc6 8.0-0 Be7 9.Bg5 h6 10.Bxf6 Bxf6 11.Qd2 0-0 12.Rad1 Qb6 13.Nd5 Qxb2 14.Nxf6+ gxf6 15.Qxh6 Qxc2
  +/=  (0.26)   Depth: 14/31   00:00:00  1153kN
6...e5 7.Nf3 Nc6 8.0-0 Be7 9.Bg5 0-0 10.Bxf6 Bxf6 11.Nd5 Be6 12.c3 a6 13.Qb3 Bxd5 14.exd5 Na5 15.Qc2 Qd7 16.Bd3
  =  (0.24)   Depth: 15/32   00:00:00  2002kN
6...e5 7.Nb3 Be7 8.0-0 Nc6 9.Qd3 0-0 10.Qg3 g6 11.Rd1 Nb4 12.Bh6 Re8 13.Rac1 Bd7 14.h3
  +/=  (0.27)   Depth: 16/32   00:00:01  3665kN
6...e5 7.Nb3 Be7 8.0-0 Nc6 9.Qd3 0-0 10.Qg3 g6 11.Bh6 Re8 12.Rad1 Qc7 13.Nb5 Qb6 14.Qf3 Bg4
  +/=  (0.29)   Depth: 17/32   00:00:01  5524kN
6...e5
  =  (0.22 !)   Depth: 18/35   00:00:02  11040kN
6...e5 7.Nb3 Be7 8.0-0 Nc6 9.Qd3 0-0 10.a3 Be6 11.Nd5 Qd7 12.Be3 Ng4 13.Rad1 Nxe3 14.Qxe3 f5 15.Nd2 Bxd5 16.exd5
  =  (0.21)   Depth: 18/35   00:00:03  14949kN
6...e5 7.Nb3 Be7 8.0-0 Nc6 9.Qd3 0-0 10.a3 Be6 11.Nd5 Qd7 12.c4 Rad8 13.Be3 Bg4 14.Nd2 Bxe2 15.Qxe2 Nxd5 16.cxd5 Nd4 17.Bxd4 exd4
  +/=  (0.26)   Depth: 19/36   00:00:04  18984kN
6...e5 7.Nb3 Be7 8.0-0 Nc6 9.Qd3 0-0 10.a3 Be6 11.Nd5 Bxd5 12.exd5 e4 13.Qb5 Ne5 14.Qxb7 Rc8 15.Nd4 Rc7 16.Qb3 Rc5 17.Bd2 Rxd5 18.Qe3 Qb6
  =  (0.25)   Depth: 20/38   00:00:07  33018kN
6...e5 7.Nb3 Be7 8.0-0 Nc6 9.Qd3 0-0 10.a3 Be6 11.Nd5 Rc8 12.c4 b5 13.Nxf6+ Bxf6 14.cxb5 Ne7 15.Bd2 d5 16.Ba5 Qd7 17.Rad1
  =  (0.24)   Depth: 21/38   00:00:13  64768kN

(Grayson, Newport, South Wales 08.05.2012)
Parent - By Hurnavich (Bronze) Date 2012-05-08 20:36
New game
rnbqkb1r/pp3ppp/3ppn2/8/3NP3/2N5/PPP1BPPP/R1BQK2R b KQkq - 0 1


Analysis by Vitruvius_1.11H_HEM_x64:

1...e5 2.Nf3 h6 3.Bc4 Nc6 4.0-0 a6 5.Bb3 Qc7 6.Be3 Be6
  +/=  (0.43)   Depth: 6   00:00:00  1kN
1...e5 2.Nf3 h6 3.Bc4 Nc6 4.0-0 a6 5.Bb3 Qc7 6.Be3 Be6
  +/=  (0.43)   Depth: 6   00:00:00  1kN
1...e5 2.Nf3 h6 3.Bc4 Nc6 4.0-0 a6 5.Bb3 Qc7 6.Be3 Be6
  +/=  (0.43)   Depth: 7   00:00:00  2kN
1...e5 2.Nf3 h6 3.Bc4 Nc6 4.0-0 a6 5.Bb3 Qc7 6.Be3 Be6
  +/=  (0.43)   Depth: 7   00:00:00  2kN
1...e5 2.Nf3 Nc6 3.0-0 h6 4.Bc4 a6 5.Bb3 Qc7 6.Be3 Be6
  +/=  (0.43)   Depth: 8   00:00:00  3kN
1...e5 2.Nf3 Nc6 3.0-0 h6 4.Bc4 a6 5.Bb3 Qc7 6.Be3 Be6
  +/=  (0.43)   Depth: 8   00:00:00  3kN
1...e5 2.Nf3 Nc6 3.0-0 h6 4.Bc4 a6 5.Bb3 Qc7 6.Be3 Be6
  +/=  (0.43)   Depth: 9   00:00:00  3kN
1...e5 2.Nf3 Nc6 3.0-0 h6 4.Bc4 a6 5.Bb3 Qc7 6.Be3 Be6
  +/=  (0.43)   Depth: 9   00:00:00  3kN
1...e5 2.Nf3 Nc6 3.0-0 h6 4.Bc4 a6 5.Bb3 Qc7 6.Be3 Be6
  +/=  (0.43)   Depth: 10   00:00:00  3kN
1...e5 2.Nf3 Nc6 3.0-0 h6 4.Bc4 a6 5.Bb3 Qc7 6.Be3 Be6
  +/=  (0.43)   Depth: 10   00:00:00  4kN
1...e5 2.Nf3 Nc6 3.0-0 h6 4.Bc4 a6 5.Bb3 Qc7 6.Be3 Be6
  +/=  (0.43)   Depth: 11   00:00:00  4kN
1...Be7
  +/=  (0.35 ++)   Depth: 11   00:00:00  162kN
1...Be7 2.0-0 0-0 3.h3 Nc6 4.Be3 d5 5.Nxc6 bxc6 6.e5 Nd7 7.Qd4 Rb8
  +/=  (0.35)   Depth: 11   00:00:00  224kN
1...Be7 2.0-0 0-0 3.h3 Nc6 4.Be3 d5 5.Nxc6 bxc6 6.e5 Nd7 7.Qd4 Rb8
  +/=  (0.35)   Depth: 11   00:00:00  232kN
1...e5 2.Nf3 Nc6 3.0-0 h6 4.Be3 Be6 5.Qd3 a6 6.Rfd1 Be7 7.Nh4 0-0 8.Nf5
  +/=  (0.40)   Depth: 12   00:00:00  783kN
1...e5 2.Nf3 Nc6 3.0-0 h6 4.Be3 Be6 5.Qd3 a6 6.Rfd1 Be7 7.Nh4 0-0 8.Nf5
  +/=  (0.40)   Depth: 12   00:00:00  792kN
1...e5 2.Nf3 Nc6 3.0-0 h6 4.Be3 Be7 5.h3 0-0 6.Nh4 a6 7.Nf5 Bxf5 8.exf5 d5
  +/=  (0.40)   Depth: 13   00:00:00  795kN
1...e5 2.Nf3 Nc6 3.0-0 h6 4.Be3 Be7 5.h3 0-0 6.Nh4 a6 7.Nf5 Bxf5 8.exf5 d5
  +/=  (0.40)   Depth: 13   00:00:00  845kN
1...e5 2.Nf3 Nc6 3.0-0 h6 4.Be3 Be6 5.Qd3 a6 6.Rfd1 Be7 7.Nh4 Nb4 8.Qd2 0-0 9.Nf5
  +/=  (0.32 ++)   Depth: 14   00:00:00  1115kN
1...e5 2.Nf3 Nc6 3.0-0 Be7 4.h3 0-0 5.Be3 a6 6.Qd2 b5 7.a3 Be6 8.Ng5
  +/=  (0.40)   Depth: 14   00:00:00  1171kN
1...e5 2.Nf3 Nc6 3.0-0 Be7 4.h3 0-0 5.Be3 a6 6.Qd2 b5 7.a3 Be6 8.Ng5
  +/=  (0.40)   Depth: 14   00:00:00  1480kN
1...e5 2.Nf3 Nc6 3.0-0 Be7 4.h3 0-0 5.Be3 a6 6.Qd2 h6 7.Nh4 Qc7 8.Nf5 Bxf5 9.exf5
  +/=  (0.32 ++)   Depth: 15   00:00:00  1723kN
1...e5 2.Nf3 Nc6 3.0-0 Be7 4.Be3 0-0 5.Qd2 h6 6.Rad1 Be6 7.a3 Rc8 8.h3 Qd7
  +/=  (0.34)   Depth: 15   00:00:00  1854kN
1...e5 2.Nf3 Nc6 3.0-0 Be7 4.Be3 0-0 5.Qd2 h6 6.Rad1 Be6 7.a3 Rc8 8.h3 Qd7
  +/=  (0.34)   Depth: 15   00:00:00  2081kN
1...e5 2.Nf3 Nc6 3.0-0 Be7 4.Be3 0-0 5.Qd2 h6 6.Rad1 Be6 7.a3 Rc8 8.h3 Qd7
  +/=  (0.26 ++)   Depth: 16   00:00:00  4007kN
1...e5 2.Nf3 Nc6 3.0-0 Be7 4.Be3 0-0 5.Qd2 h6 6.Nh4 Be6 7.Nf5 Bxf5 8.exf5 Qd7 9.Qd3 d5
  +/=  (0.30)   Depth: 16   00:00:00  4107kN
1...e5 2.Nf3 Nc6 3.0-0 Be7 4.Be3 0-0 5.Qd2 h6 6.Nh4 Be6 7.Nf5 Bxf5 8.exf5 Qd7 9.Qd3 d5
  +/=  (0.30)   Depth: 16   00:00:00  4415kN
1...e5 2.Nb3 Nc6 3.0-0 Be6 4.f4 exf4 5.Bxf4 Be7 6.Qd2 0-0 7.Rad1 Ne5 8.Nd4 Qb6 9.b3 Rac8 10.Qe3 Rxc3 11.Qxc3 Nxe4
  =  (0.23)   Depth: 17   00:00:01  10109kN
1...e5 2.Nb3 Nc6 3.0-0 Be6 4.f4 exf4 5.Bxf4 Be7 6.Qd2 0-0 7.Rad1 Ne5 8.Nd4 Qb6 9.b3 Rac8 10.Qe3 Rxc3 11.Qxc3 Nxe4
  =  (0.23)   Depth: 17   00:00:01  10599kN
1...e5 2.Nb3 Nc6 3.0-0 Be6 4.f4 exf4 5.Bxf4 Be7 6.Be3 0-0 7.Rb1 Ne5 8.Nd4 a6 9.Nf5 Rc8 10.Qd2 Qd7
  +/=  (0.31 --)   Depth: 18   00:00:03  27460kN
1...e5 2.Nb3 Nc6 3.0-0 Be7 4.Qd3 Nb4 5.Qb5+ Nc6 6.Be3 a6 7.Qd3 Be6 8.a3 d5 9.Nc5 dxe4 10.Qxd8+ Rxd8 11.Nxe6 fxe6 12.Rfe1 Nd4
  =  (0.24)   Depth: 18   00:00:03  30549kN
1...e5 2.Nb3 Nc6 3.0-0 Be7 4.Qd3 Nb4 5.Qb5+ Nc6 6.Be3 a6 7.Qd3 Be6 8.a3 d5 9.Nc5 dxe4 10.Qxd8+ Rxd8 11.Nxe6 fxe6 12.Rfe1 Nd4
  =  (0.24)   Depth: 18   00:00:03  30550kN
1...e5 2.Nb3 Nc6 3.0-0 Be7 4.Qd3 Nb4 5.Qb5+ Nc6 6.Be3 a6 7.Qd3 Be6 8.a3 0-0 9.Nd5 Bxd5 10.exd5 e4 11.Qd2 Ne5 12.h3 Qc7
  +/=  (0.29)   Depth: 19   00:00:03  33802kN
1...e5 2.Nb3 Nc6 3.0-0 Be7 4.Qd3 Nb4 5.Qb5+ Nc6 6.Be3 a6 7.Qd3 Be6 8.a3 0-0 9.Nd5 Bxd5 10.exd5 e4 11.Qd2 Ne5 12.h3 Qc7
  +/=  (0.29)   Depth: 19   00:00:04  35374kN
1...e5 2.Nb3 Nc6 3.0-0 Be7 4.Qd3 Nb4 5.Qb5+ Nc6 6.Be3 a6 7.Qd3 Be6 8.a3 0-0 9.Nd5 Bxd5 10.exd5 e4 11.Qd2 Ne5 12.h3 Qc7
  +/=  (0.29)   Depth: 20   00:00:05  44767kN
1...e5 2.Nb3 Nc6 3.0-0 Be7 4.Qd3 Nb4 5.Qb5+ Nc6 6.Be3 a6 7.Qd3 Be6 8.a3 0-0 9.Nd5 Bxd5 10.exd5 e4 11.Qd2 Ne5 12.h3 Qc7
  +/=  (0.29)   Depth: 20   00:00:07  65014kN
1...e5 2.Nb3 Nc6 3.0-0 Be7 4.Qd3 Nb4 5.Qb5+ Nc6 6.Be3 a6 7.Qd3 Be6 8.a3 0-0 9.Nd5 Bxd5 10.exd5 e4 11.Qd2 Ne5 12.h3 Qc7
  +/=  (0.29)   Depth: 21   00:00:09  84551kN
1...e5 2.Nb3 Nc6 3.0-0 Be7 4.Qd3 Nb4 5.Qb5+ Nc6 6.Be3 a6 7.Qd3 Be6 8.a3 0-0 9.Nd5 Bxd5 10.exd5 e4 11.Qd2 Ne5 12.h3 Qc7
  +/=  (0.29)   Depth: 21   00:00:31  295mN
1...e5 2.Nb3 Nc6 3.0-0 Be7 4.Qd3 Nb4 5.Qb5+ Nc6 6.Be3 0-0 7.a3 h6 8.Qd3 Be6 9.Nd5 Qd7 10.h3 Rac8 11.Rad1 Rfd8 12.Nxe7+ Qxe7 13.Nd2 d5
  +/=  (0.31)   Depth: 22   00:00:40  380mN
1...a6 2.0-0 Nbd7 3.Qd3 b6 4.Qe3 Bb7 5.Rd1 Qc7 6.a3 b5 7.b4 e5 8.Nf5 g6 9.Ng3 Rc8 10.Bb2 Qb6 11.a4 Qxe3 12.fxe3 Bh6 13.axb5 axb5 14.Nxb5 Bxe3+ 15.Kh1
  =  (0.24)   Depth: 22   00:00:58  547mN
1...a6 2.0-0 Nbd7 3.Qd3 b6 4.Qe3 Bb7 5.Rd1 Qc7 6.a3 b5 7.b4 e5 8.Nf5 g6 9.Ng3 Rc8 10.Bb2 Qb6 11.a4 Qxe3 12.fxe3 Bh6 13.axb5 axb5 14.Nxb5 Bxe3+ 15.Kh1
  =  (0.24)   Depth: 22   00:01:00  565mN
1...a6 2.0-0 Nbd7 3.Qd3 b6 4.Qe3 Bb7 5.Rd1 Qc7 6.a3 b5 7.b4 e5 8.Nf5 g6 9.Ng3 Rc8 10.Bb2 Qb6 11.a4 Qxe3 12.fxe3 Bh6 13.axb5 axb5 14.Nxb5 Bxe3+ 15.Kh1
  =  (0.24)   Depth: 23   00:01:19  751mN
1...a6 2.0-0 Nbd7 3.Qd3 b6 4.Qe3 Bb7 5.Rd1 Qc7 6.a3 b5 7.b4 e5 8.Nf5 g6 9.Ng3 Rc8 10.Bb2 Qb6 11.a4 Qxe3 12.fxe3 Bh6 13.axb5 axb5 14.Nxb5 Bxe3+ 15.Kh1
  =  (0.24)   Depth: 23   00:01:23  788mN

(Hurnavich, 08.05.2012)
Parent - By Barnard (Bronze) Date 2012-05-08 08:56
agree about the hardware...Houdini played very serious blunders at moves 15 and 17 that made him losse the game...

15...h6 is a blunder,but 17...Nc6 is a very big blunder,directly lossing the game
Parent - - By shrapnel (***) Date 2012-05-09 13:59
Well, first of all you handicap Houdini by forcing it to play a plain silly move like 5....e6 from that weird book ( 5.... a6 is the generally accepted move, according to Live Book, Fritz 13). I had my Houdini 2.0c Pro x64 Engine running side-by-side with Live Book ; it took one look at 5....e6 for a microsecond, sneered,:grin: and moved on to consider 5...g6 and 5.....a6.
Coming to the 15th move (needless to say many questionable moves in between) , my Houdini doesn't even LOOK at the ridiculous 15.....h6 and makes the move 15....Bh4.
I'm getting tired of people trying to put Houdini down, while the plain simple FACT is that no other Engine comes even close to matching the Power and Strength of Houdini.
Oh well, I suppose many people can't stomach this plain and simple truth and keep trying to find faults in the magnificent Houdini engine:roll::roll:
Parent - - By Barnard (Bronze) Date 2012-05-09 14:09
i think that all the mistakes Houdini made was due to the hardware used,a weak hardware.and that explains all the weak and strange moves
Parent - - By Josef (**) Date 2012-05-09 22:18
I don't agree.

First, the Core 2 Duo single is about 1/12 of the Core i7 quad, resulting in < 200 Elo difference (correct me if I'm wrong). Remains 3100 Elo for Houdini on my computer. I believe the blunders are way beyond that strength, maybe even in the 2000 range.

Second, that Houdini does not play the blunder moves on another machine says nothing, one has to check in multi variant mode if there is a noteable difference in the moves rating.

Generally, I don't belive you gain that much quality from an office PC to a highend machine, so I wouldn't spend money in one. Only Komodos overproportional gain changes my view a bit, but from my observation Houdini does not keep up here and achives only mean gain.
Parent - - By Barnard (Bronze) Date 2012-05-09 22:20
so why you think that Houdini played soo much mistakes only in your machine?due to what?
Parent - - By Josef (**) Date 2012-05-09 23:28
2 possible reasons.

1. the other machine plays another move by chance (which is likely especially when changing from single- to multicore). Any other reasonable move will not be a blunder, so the engine avoids it.

2. the higher calculation power leads the engine to see the blunder. As said one can only check in multivariant mode that the blunder move distinctly drops. But since the engine doesn't understand this position aside from tactical search, similar positions will arise eventually which take 2 moves deeper calculation to see something tactically, and the same problem occurs on the faster hardware. Of course such positions occur rarely, and if, there is only a small chance for playing a blunder even without understanding, so that doesn't hurt statistical strength of the engine. But still happens, when you search for it.

You also have to consider the continuous pace of processor speedup. The Core 2 was top some 5 years ago, nobody would have said then it's a weak machine for chess. In another 5 years computers will have increased same step (if that happens, but let's assume) and some people surely will write the Core i7 is weak hardware which is prone to gross blunders. So, on which ground should it play strong now and weak then?
Parent - - By Barnard (Bronze) Date 2012-05-09 23:43
my machine is a core 2 duo,so i can reproduce the same analisis,and is the laptp that im using to play the WBCCC 2 tournament,that few days ago my laptop was broken and i had even an older laptop,but now i have mine again

what i dont understand is why you handicaped Houdini with the openings book
Parent - - By Josef (**) Date 2012-05-10 20:00
My private machine is a 6 year old Pentium D. It's still ok for the internet, but not so good for chess, since at same Ghz achieves only 1/2 the speed of the Core 2 Duo. It also burns a lot of energy (180 watt when calculating) so leaving the computer on for engine matches is no option. Today I believe the Pentium D is a bit misengineered, but I didn't realize that when buying. Payed 1000 € for it in 2006. Anyhow, no single hw issue within 6 years (although some viruses).

Concerning opening book I believe more beautiful games arise when 2 engines are close in strength, so I help the weaker one with a book.
Parent - - By Barnard (Bronze) Date 2012-05-10 20:32
but yu cant help the weaker engine,just because the weaker is weaker is weaker...

is the same that if in a race,a slower racer you make he start running 20 meters after the others...that isnt a just race
Parent - - By Josef (**) Date 2012-05-10 20:39
I visited a chess club for some time years ago, and most players there offered me material. Because they were better and wanted to play an interesting game.
Parent - By Barnard (Bronze) Date 2012-05-10 20:43
your example is different;your example some players help you to improve your skills,you only handicaped one engine...one example is totally different from the other
Parent - - By leavenfish (***) Date 2012-05-11 02:33
Likewise, when you are looking for a good wine, you don't pay for the one that made it from vineyard to table the quickest.

I mean, the clock is part of the game, but a chess game isn't decided by who can slap the clock the quickest...it's about the quality of the moves.
Parent - By Barnard (Bronze) Date 2012-05-11 18:12
my friend,agree partially with you...the game is decided by who play the better chess and who ran out of time before
Parent - - By ernest (****) Date 2012-05-10 18:39

> the Core 2 Duo single is about 1/12 of the Core i7 quad


Unless you consider the GHz of the Core i7 as double the GHz of the Core 2 Duo, this is nonsense...
Parent - - By Josef (**) Date 2012-05-10 19:49
Single nodespeed on 3 Ghz Core 2 Duo is about 350 (Komodo), 1270 (Houdini)
Single nodespeed on Core i7 @ 2.80 Ghz 3830 Mhz (whatever this means) is about 1360 (Komodo), 3070 (Houdini)

So where is the nonsense?
Komodo seems to gain more.
Parent - - By ernest (****) Date 2012-05-12 18:50

> So where is the nonsense?


Something is not right with your numbers!
How did you get them?

Even if the Core i7 is 28% faster than the Core 2 Duo (3830 MHz compared to 3000 MHz), you cannot get x3.9 (Komodo) or x2.4 (Houdini) in nodespeed.
Parent - By Josef (**) Date 2012-05-14 05:13
The numbers for Core 2 Duo are from my own test, see the game of this thread.
The numbers for Core i7 are from this thread: http://rybkaforum.net/cgi-bin/rybkaforum/topic_show.pl?tid=23839

Anyway, if the factor should be less, it even more supports my point that it doesn't pay off to switch from Core 2 Duo to Core i7, especially when using single core Komodo.
Parent - - By NATIONAL12 (Gold) Date 2012-05-09 22:59
that microsecond was to long in this game and why houdini played it.To me and i have stated this many times on forum these games are micky mouse.
Parent - By Uly (Gold) Date 2012-05-10 04:28
Parent - By shrapnel (***) Date 2012-05-10 12:00
Micky mouse games just about sums it up.... and just look at the completely misleading title of this topic:mad:
Parent - By Pia (****) Date 2012-05-10 22:26

>my Houdini doesn't even LOOK at the ridiculous 15.....h6


Engines tend to make weaker moves while playing by using hash,
or previous analysis, to save time. It's good example of this I think.
Up Topic The Rybka Lounge / Computer Chess / Komodo crushes Houdini in 25 moves from standard opening

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill