hard endgame test position

Discussion of anything and everything relating to chess playing software and machines.

Moderator: Ras

User avatar
smirobth
Posts: 2307
Joined: Wed Mar 08, 2006 8:41 pm
Location: Brownsville Texas USA

hard endgame test position

Post by smirobth »

[d] 8/5pk1/8/3bP1pp/3P1p2/7P/3B1K2/8 b - - 0 1

1...h4! wins. Most programs seem to want to play 1...Kg6 after which 2.h4 draws. Other moves besides 1...h4 also only draw.
- Robin Smith
Top
swami
Posts: 6663
Joined: Thu Mar 09, 2006 4:21 am

Re: hard endgame test position

Post by swami »

Hi Robin,

On contrary,it seems most of the programs find it and they choose to play f6 rather than Kg6,and some programs ply h4 with greater than +1.3 evaluation score,I don't know if that's a good score enough to win in particular position.


FEN: 8/5pk1/8/3bP1pp/3P1p2/7P/3B1K2/8 w - - 0 1

Code: Select all

Green Light Chess 301.2.2:
   1	00:00	           1	1	+0.68	Ba8
   1	00:00	           3	3	+0.69	Bb7
   1	00:00	           5	5	+0.90	Be6
   1	00:00	          20	20	+0.93	Kg6
   1	00:00	          26	26	+1.00	f6
   1	00:00	          30	30	+1.00	f6
   2	00:00	          79	79	+1.02	f6 2. exf6+ Kxf6 3. Bc3
   2	00:00	         149	149	+1.02	f6 2. exf6+ Kxf6 3. Bc3
   3	00:00	         451	451	+1.03	f6 2. exf6+ Kxf6 3. Ke2 Kf5
   3	00:00	         558	558	+1.03	f6 2. exf6+ Kxf6 3. Ke2 Kf5
   4	00:00	       1.485	1.485	+0.74	f6 2. exf6+ Kxf6 3. h4 Kg6 4. hxg5 Kxg5
   4	00:00	       2.945	2.945	+0.96	h4 2. Bb4 f6 3. exf6+ Kxf6 4. Bc5
   4	00:00	       2.949	2.949	+0.96	h4 2. Bb4 f6 3. exf6+ Kxf6 4. Bc5
   5	00:00	       3.812	3.812	+0.97	h4 2. Bb4 f6 3. exf6+ Kxf6 4. Bd6 Kf5
   5	00:00	       5.430	5.430	+0.97	h4 2. Bb4 f6 3. exf6+ Kxf6 4. Bd6 Kf5
   6	00:00	       7.602	7.602	+0.83	h4 2. Bb4 Kg6 3. Be7 Be6 4. Kg2
   6	00:01	      14.315	477.166	+0.83	h4 2. Bb4 Kg6 3. Be7 Be6 4. Kg2
   7	00:01	      16.948	564.933	+0.85	h4 2. Bb4 Kg6 3. Be7 Be6 4. Kg2 Bd5+ 5. Kf2
   7	00:01	      27.328	683.200	+0.85	h4 2. Bb4 Kg6 3. Be7 Be6 4. Kg2 Bd5+ 5. Kf2
   8	00:01	      38.937	648.950	+0.82	h4 2. Bb4 f5 3. Be7 Kh6 4. Bf6 Kg6 5. Ke2
   8	00:01	      64.246	917.800	+0.82	h4 2. Bb4 f5 3. Be7 Kh6 4. Bf6 Kg6 5. Ke2
   9	00:01	      77.059	770.590	+0.86	h4 2. Bb4 f5 3. Be7 Kh6 4. Bf6 Kh5 5. Ke2 g4
   9	00:01	      92.556	771.300	+0.86	h4 2. Bb4 f5 3. Be7 Kh6 4. Bf6 Kh5 5. Ke2 g4
  10	00:01	     175.661	548.940	+0.88	h4 2. Ba5 f5 3. Bd8 Kg6 4. Kf1 f3 5. Kf2 f4 6. Bf6
  10	00:01	     196.730	578.617	+0.88	h4 2. Ba5 f5 3. Bd8 Kg6 4. Kf1 f3 5. Kf2 f4 6. Bf6
  11	00:01	     395.203	470.479	+1.01	h4 2. Ba5 f5 3. exf6+ Kxf6 4. Bd8+ Kf5 5. Kg1 g4 6. Bxh4 gxh3 7. Kh2 Kg4 8. Bf6
  11	00:01	     419.624	466.248	+1.01	h4 2. Ba5 f5 3. exf6+ Kxf6 4. Bd8+ Kf5 5. Kg1 g4 6. Bxh4 gxh3 7. Kh2 Kg4 8. Bf6
  12	00:01	     863.858	543.306	+1.10	h4 2. Kg1 f5 3. Be1 Kf7 4. Bc3 f3 5. Ba5 Kg6 {ht} 6. Bd8
  12	00:01	     897.740	537.568	+1.10	h4 2. Kg1 f5 3. Be1 Kf7 4. Bc3 f3 5. Ba5 Kg6 {ht} 6. Bd8
  13	00:02	   1.296.323	617.296	+1.11	h4 2. Kg1 f5 3. exf6+ Kxf6 4. Kf2 Kf5 5. Kg1 g4 6. hxg4+ Kxg4 7. Kh2 f3 8. Be1 h3 9. Bf2 Be4
  13	00:02	   1.360.191	618.268	+1.11	h4 2. Kg1 f5 3. exf6+ Kxf6 4. Kf2 Kf5 5. Kg1 g4 6. hxg4+ Kxg4 7. Kh2 f3 8. Be1 h3 9. Bf2 Be4
  14	00:03	   2.678.929	750.400	+1.24	h4 2. Kg1 f5 3. exf6+ Kxf6 4. Kf2 Kf5 5. Kg1 g4 6. hxg4+ Kxg4 7. Kh2 f3 8. Be1 h3 9. Bf2 Be4 10. Be3
  14	00:03	   2.762.709	746.678	+1.24	h4 2. Kg1 f5 3. exf6+ Kxf6 4. Kf2 Kf5 5. Kg1 g4 6. hxg4+ Kxg4 7. Kh2 f3 8. Be1 h3 9. Bf2 Be4 10. Be3
  15	00:04	   3.738.091	817.963	+1.21	h4 2. Kg1 f5 3. exf6+ Kxf6 4. Kf2 Kf5 5. Kg1 g4 6. hxg4+ Kxg4 7. Bb4 h3 8. Kh2 f3 9. Be1 Be4 10. Bf2 Bd5
  15	00:05	   4.011.478	832.256	+1.21	h4 2. Kg1 f5 3. exf6+ Kxf6 4. Kf2 Kf5 5. Kg1 g4 6. hxg4+ Kxg4 7. Bb4 h3 8. Kh2 f3 9. Be1 Be4 10. Bf2 Bd5
  16	00:06	   4.986.364	853.829	+1.21	h4 2. Kg1 f5 3. exf6+ Kxf6 4. Ba5 Kf5 5. Kf2 g4 6. hxg4+ Kxg4 7. Kg1 h3 8. Kh2 f3 9. Be1 {ht} Be4
  16	00:06	   5.338.488	842.032	+1.21	h4 2. Kg1 f5 3. exf6+ Kxf6 4. Ba5 Kf5 5. Kf2 g4 6. hxg4+ Kxg4 7. Kg1 h3 8. Kh2 f3 9. Be1 {ht} Be4
  17	00:17	  16.448.080	948.562	+1.45	h4 2. Kg1 Kg6 3. Kf2 Kf5 4. Bb4 f3 5. Kg1 g4 6. hxg4+ Kxg4 7. Be1 h3 8. Kh2 Bb3 9. Bf2 Be6 {ht} 10. Be3
  17	00:18	  17.089.742	944.706	+1.45	h4 2. Kg1 Kg6 3. Kf2 Kf5 4. Bb4 f3 5. Kg1 g4 6. hxg4+ Kxg4 7. Be1 h3 8. Kh2 Bb3 9. Bf2 Be6 {ht} 10. Be3
  18	00:22	  21.004.734	950.870	+1.43	h4 2. Kg1 Kg6 3. Kf2 Kf5 4. Bb4 f3 5. Be1 g4 6. hxg4+ Kxg4 7. Kg1 {ht} Be6
  18	00:24	  22.872.509	953.021	+1.43	h4 2. Kg1 Kg6 3. Kf2 Kf5 4. Bb4 f3 5. Be1 g4 6. hxg4+ Kxg4 7. Kg1 {ht} Be6
  19	00:29	  27.470.568	981.091	+1.40	h4 2. Kg1 Kg6 3. Kf2 Kf5 4. Bb4 f3 5. Be1 Kg6 6. Bd2 Kh5 7. Kg1 Be6 8. Kh2 Bxh3 9. Kxh3 g4+ 10. Kh2 g3+ 11. Kg1 f2+ 12. Kf1 h3
  19	00:33	  31.764.193	962.551	+1.40	h4 2. Kg1 Kg6 3. Kf2 Kf5 4. Bb4 f3 5. Be1 Kg6 6. Bd2 Kh5 7. Kg1 Be6 8. Kh2 Bxh3 9. Kxh3 g4+ 10. Kh2 g3+ 11. Kg1 f2+ 12. Kf1 h3
  20	00:49	  46.017.691	939.136	+1.43	h4 2. Kg1 Kg6 3. Kf2 Kf5 4. Bb4 f3 5. Be1 Kg6 6. Bd2 Kh5 7. Kg1 Bc4 8. Kf2 g4 9. hxg4+ Kxg4 10. Be3 Be6 11. Bc1 {ht} h3
  20	00:55	  51.837.531	942.500	+1.43	h4 2. Kg1 Kg6 3. Kf2 Kf5 4. Bb4 f3 5. Be1 Kg6 6. Bd2 Kh5 7. Kg1 Bc4 8. Kf2 g4 9. hxg4+ Kxg4 10. Be3 Be6 11. Bc1 {ht} h3
  21	01:22	  78.605.082	970.433	+1.43	h4 2. Kg1 Kg6 3. Kf2 Kf5 4. Bb4 f3 5. Be1 Kg6 6. Bd2 Kh5 7. Kg1 Bc4 8. Kf2 g4 9. Bf4 Bd5 10. hxg4+ Kxg4 11. Bh2 h3 12. Ke3 Be6
Top
User avatar
smirobth
Posts: 2307
Joined: Wed Mar 08, 2006 8:41 pm
Location: Brownsville Texas USA

Re: hard endgame test position

Post by smirobth »

swami wrote:Hi Robin,

On contrary,it seems most of the programs find it and they choose to play f6 rather than Kg6,and some programs ply h4 with greater than +1.3 evaluation score,I don't know if that's a good score enough to win in particular position. FEN: 8/5pk1/8/3bP1pp/3P1p2/7P/3B1K2/8 w - - 0 1

Code: Select all

Green Light Chess 301.2.2:
Interesting. Perhaps I should have said that Rybka, Shredder, Fritz and Hiarcs like Kg6, and all take more than 10 minutes to find h4. h4 does definately win. You can't go by just the +1.3. Push any of the programs deeper and they will all eventually start showing very high scores.
Last edited by smirobth on Wed Aug 15, 2007 6:59 am, edited 1 time in total.
- Robin Smith
Top
swami
Posts: 6663
Joined: Thu Mar 09, 2006 4:21 am

Re: hard endgame test position

Post by swami »

smirobth wrote:
swami wrote:FEN: 8/5pk1/8/3bP1pp/3P1p2/7P/3B1K2/8 w - - 0 1

Code: Select all

Green Light Chess 301.2.2:
Interesting. Rybka, Shredder, Fritz and Hiarcs all take more than 10 minutes.
It is rated over 2600, not a bad program compared to strongest ones.
Top
User avatar
smirobth
Posts: 2307
Joined: Wed Mar 08, 2006 8:41 pm
Location: Brownsville Texas USA

Re: hard endgame test position

Post by smirobth »

swami wrote:
smirobth wrote:
swami wrote:FEN: 8/5pk1/8/3bP1pp/3P1p2/7P/3B1K2/8 w - - 0 1

Code: Select all

Green Light Chess 301.2.2:
Interesting. Rybka, Shredder, Fritz and Hiarcs all take more than 10 minutes.
It is rated over 2600, not a bad program compared to strongest ones.
I have not used Green Light Chess. Maybe I should check it out.
- Robin Smith
Top
BBauer
Posts: 658
Joined: Wed Mar 08, 2006 8:58 pm

Re: hard endgame test position

Post by BBauer »

Again a very interesting position.
Programs have great problems playing endgames with bishops of opposite coulors. In the book of Awerbach I found an endgame where the program plays the rihgt move in 0 sec, but fails to find the right continuation. Often 2 more pawns are not enough to win, but the program is lured by a high score.

To your position:
I found

1. ...h4 2.Bb4 Kg6
3.Be7 Kf5 4.Bb4 Ke4
5.Kg1 Kf3 6.Be7 g4
7.hxg4 h3 8.Kh2 Ke2
9.Bh4 f3 10.Kxh3 Be6
11.Bg3 f2 12.Bxf2 Kxf2
13.Kh4 0-1

Do you agree?

kind regards
Bernhard
Top
BBauer
Posts: 658
Joined: Wed Mar 08, 2006 8:58 pm

Re: hard endgame test position

Post by BBauer »

Here is output of Fruit2.3 on one proc of a AMD X2 5200+

Code: Select all

FEN: 8/5pk1/8/3bP1pp/3P1p2/7P/3B1K2/8 b - - 0 1 

Fruit 2.3:
  18	00:02	   2.568.286	1.229.096	+1,04	f7f6 h3h4 Kg7g6 Kf2g1 f4f3 e5xf6 g5g4 Kg1f2 Kg6xf6 Bd2b4 Kf6f5 Bb4d6 Kf5e4 Bd6e5 Bd5c4 Be5f6 Ke4d5 Bf6e5 Bc4d3
  19	00:04	   5.035.370	1.226.667	+1,06	f7f6 h3h4 Kg7g6 e5xf6 Kg6xf6 Kf2f1 Kf6f5 h4xg5 Kf5xg5 Bd2e1 h5h4 Kf1g1 Kg5g4 Kg1h2 f4f3 Be1f2 Bd5e6 Bf2e3 Be6c4 Be3f2 h4h3 Bf2g3
  20	00:05	   6.570.732	1.276.347	+1,05	f7f6 h3h4 Kg7g6 e5xf6 Kg6xf6 Kf2f1 Kf6f5 h4xg5 Kf5xg5 Bd2e1 Bd5c4+ Kf1g1 Kg5g4 Kg1g2 h5h4 Kg2h2 f4f3 Be1f2 Bc4d5 Bf2e1 Bd5f7 Be1f2 Bf7c4 Bf2e3
  21	00:06	   8.954.272	1.375.000	+1,06	f7f6 h3h4 Kg7g6 e5xf6 Kg6xf6 Kf2f1 Kf6f5 h4xg5 Kf5xg5 Bd2e1 Bd5c4+ Kf1g1 Kg5f5 Kg1g2 Kf5e4 Be1f2 Bc4e6 Kg2h2 f4f3 Kh2g3 Be6c4 Kg3h4 Bc4f7 Bf2g1 Ke4d5
  21	00:09	  12.734.527	1.408.750	+1,16	Kg7g6 h3h4 Kg6f5 h4xg5 Kf5e4 Kf2f1 h5h4 Kf1g1 h4h3 Kg1h2 f4f3 Bd2e1 Ke4e3 Kh2xh3 Ke3e2 Be1g3 f3f2 Bg3xf2 Ke2xf2 Kh3g4 Kf2e3 g5g6 f7xg6 e5e6
  22	00:11	  16.378.721	1.394.545	+1,14	Kg7g6 h3h4 Kg6f5 h4xg5 Kf5e4 Kf2g1 f4f3 Bd2e1 Ke4xd4 Be1g3 Kd4e3 Kg1f1 Bd5e4 Kf1e1 Be4f5 Bg3h4 Bf5e6 Bh4f2+ Ke3f4 Bf2b6 h5h4 Ke1f2 h4h3 Bb6e3+ Kf4xe5 Kf2xf3
  23	00:13	  19.545.197	1.421.538	+1,16	Kg7g6 h3h4 Kg6f5 h4xg5 Kf5e4 Kf2g1 f4f3 Bd2e1 Ke4xd4 Be1g3 Kd4e3 Kg1f1 Bd5e4 Kf1e1 Be4f5 Bg3h4 Bf5e6 Ke1f1 Be6c4+ Kf1e1 Bc4d5 Bh4f2+ Ke3f4 Bf2b6 h5h4 Ke1f2 Kf4xe5
  24	00:22	  32.066.090	1.453.636	+1,37	Kg7g6 h3h4 Kg6f5 h4xg5 Kf5e4 Kf2g1 f4f3 Bd2e1 Ke4xd4 Be1g3 Kd4e4 Kg1h2 Ke4f5 Bg3f2 Kf5xg5 Kh2h3 Kg5f5 Kh3h4 Kf5xe5 Kh4xh5 f7f6 Bf2g3+ Ke5e4 Kh5g6 f6f5 Kg6g5
  25	00:26	  38.870.480	1.467.308	+1,32	Kg7g6 h3h4 Kg6f5 h4xg5 Kf5e4 Kf2g1 f4f3 Bd2e1 Ke4xd4 Be1g3 Kd4e4 Kg1h2 Ke4f5 Bg3f2 Kf5xg5 Kh2h3 Bd5e4 Bf2e3+ Kg5f5 Kh3h4 Be4d3 Kh4xh5 Kf5e4 Be3c5 Ke4d5 Bc5e3 Kd5xe5
  26	00:34	  51.117.138	1.477.353	+1,32	Kg7g6 h3h4 Kg6f5 h4xg5 Kf5e4 Kf2g1 f4f3 Bd2e1 Ke4xd4 Be1g3 Kd4e4 Kg1h2 Ke4f5 Bg3f2 Kf5xg5 Kh2h3 Bd5c6 Bf2e3+ Kg5f5 Kh3h4 Kf5e4 Be3b6 Ke4xe5 Kh4xh5 Bc6d7 Bb6c5 Bd7e6 Kh5h4
  27	00:46	  69.922.873	1.492.391	+1,27	Kg7g6 h3h4 Kg6f5 h4xg5 Kf5e4 Kf2g1 f4f3 Bd2e1 Ke4xd4 Be1g3 Kd4e4 Kg1h2 Ke4f5 Bg3f2 Kf5xg5 Kh2h3 Kg5f5 Kh3h4 Kf5xe5 Kh4xh5 f7f6 Kh5g4 f6f5+ Kg4g5 f5f4 Bf2c5 Ke5e4 Kg5h4 Bd5c4
  28	01:25	 126.619.126	1.486.000	+1,47	Kg7g6 h3h4 Kg6f5 h4xg5 Kf5e4 Kf2g1 f4f3 Bd2e1 Ke4xd4 Be1g3 Kd4e4 Kg1h2 Ke4f5 Kh2g1 Kf5xg5 Kg1f1 h5h4 Bg3h2 h4h3 Kf1f2 Kg5f5 Bh2g3 Kf5e4 Bg3h2 Bd5e6 Bh2g3 Be6d7 Kf2f1 Bd7b5+ Kf1f2
  29	02:09	 191.554.882	1.481.318	+1,49	Kg7g6 h3h4 Kg6f5 h4xg5 Kf5e4 Kf2g1 f4f3 Bd2e1 Ke4xd4 Be1g3 Kd4e4 Kg1h2 Ke4f5 Kh2g1 Kf5xg5 Kg1f1 h5h4 Bg3h2 Kg5f5 Kf1f2 Kf5e4 Kf2f1 Ke4e3 Bh2g1+ Ke3f4 Bg1h2+ Kf4e4 Kf1e1 Bd5c4 Ke1f2 h4h3 Bh2g3
  29	05:07	 432.197.777	1.408.727	+2,15	h5h4 Bd2e1 Kg7g6 Kf2g1 Kg6f5 Be1a5 Kf5e4 e5e6 f7xe6 Ba5d8 g5g4 h3xg4 h4h3 g4g5 f4f3 g5g6 e6e5 Bd8h4 e5xd4 g6g7 Bd5c4 Bh4e1 d4d3 Be1d2 Bc4d5 Bd2g5 Ke4f5 Bg5e3 f3f2+ Kg1xf2 h3h2 g7g8Q Bd5xg8
  30	09:27	 789.262.924	1.391.663	+4,05	h5h4 Bd2e1 Kg7g6 Kf2g1 Kg6f5 Be1b4 Kf5e4 Bb4d6 Ke4xd4 Bd6c7 Kd4d3 Kg1f1 Bd5e6 Bc7d8 Be6xh3+ Kf1g1 g5g4 Bd8xh4 g4g3 Kg1h1 Kd3e4 Kh1g1 Bh3e6 Kg1f1 Ke4xe5 Bh4d8 Ke5f5 Bd8c7 Be6d5 Kf1g1 f7f6 Bc7d6
  31	14:21	1.198.090.997	1.391.033	+4,10	h5h4 Kf2g1 Kg7g6 Bd2b4 Kg6f5 Bb4e7 g5g4 h3xg4+ Kf5xg4 e5e6 f7xe6 Kg1h2 f4f3 Kh2g1 h4h3 Be7d6 Kg4f5 Kg1f2 Kf5e4 Bd6e5 Bd5c6 Be5h2 Ke4xd4 Bh2g1 Kd4d3 Bg1h2 Kd3e4 Kf2g3 Bc6b5 Bh2g1 Bb5f1 Kg3f2 Bf1c4 Bg1h2
So fruit2.3 with tablebases finds the solution in 5:07 min
kind regards
Bernhard
Top
User avatar
tiger
Posts: 819
Joined: Sat Mar 11, 2006 3:15 am
Location: Guadeloupe (french caribbean island)

Re: hard endgame test position

Post by tiger »

smirobth wrote:[d] 8/5pk1/8/3bP1pp/3P1p2/7P/3B1K2/8 b - - 0 1

1...h4! wins. Most programs seem to want to play 1...Kg6 after which 2.h4 draws. Other moves besides 1...h4 also only draw.

I don't know if it's good enough to convince the experts, but here is Chess Tiger 2007.1 output on P3-933MHz (1 core), 48Mb hash tables, no endgame database:

Code: Select all

Time         Score Depth Nodes    PV
00:00:00.4	1,15	10	100699	f6 h4 fxe5 dxe5 Kg6 Ke1 Kf5 hxg5 Kxe5 
00:00:00.5	1,13	11	147948	f6 h4 fxe5 dxe5 Kg6 Bc1 Kf5 hxg5 Kxg5 Bd2 h4 
00:00:00.9	1,18	12	265160	f6 h4 fxe5 dxe5 Kg6 Bc1 Kf5 hxg5 Kxg5 Ba3 Kf5 Bd6 h4 
00:00:01.4	1,18	13	457854	f6 h4 fxe5 dxe5 Kg6 Bc1 Kf5 hxg5 Kxg5 Ba3 Kf5 Bd6 h4 
00:00:02.5	1,25	14	919654	f6 h4 fxe5 dxe5 Kg6 Ke1 gxh4 Bxf4 Kf5 Bh2 h3 Kf2 h4 
00:00:03.8	1,32	15	1446892	f6 h4 fxe5 dxe5 Kg6 Bc1 Kf5 hxg5 Kxg5 Ba3 Kf5 Bd6 h4 
00:00:06.4	1,28	16	2458799	f6 h4 fxe5 dxe5 Kg6 Kg1 gxh4 Bxf4 Kf5 Bh2 h3 Bg3 h4 Bxh4 Kxe5 Bg3+ Ke4 
00:00:11.7	1,36	17	4624637	f6 h4 fxe5 dxe5 Kg6 Kg1 Kf5 hxg5 Kxg5 Bc3 h4 Bd4 Kf5 Bf2 Kxe5 Bxh4 Ke4 
00:00:15.5	1,36	18	6195282	f6 h4 fxe5 dxe5 Kg6 Kf1 gxh4 Bxf4 Kf5 Bh2 h3 Kf2 h4 Ke3 Kg4 Kd4 Be6 
00:00:22.7	1,40	19	9201278	f6 h4 fxe5 dxe5 Kg6 Kf1 Kf5 hxg5 Kxg5 Bb4 Kf5 Bc3 h4 Kf2 h3 Kg1 f3 Bd4 Ke4 Bc3 Ke3 
00:00:40.8	1,38	20	16679159	f6 h4 fxe5 dxe5 Kg6 Kf1 Kf5 hxg5 Kxg5 Bb4 Kf5 Bc3 h4 Kf2 h3 Kg1 f3 Bd4 Ke4 Bc3 Kd3 Be1 Ke3 
00:01:18.6	1,41	21	32443485	f6 h4 fxe5 dxe5 Kg6 Kf1 Kf5 hxg5 Kxg5 Bb4 Kf5 Bc3 Ke4 Kf2 f3 Kg3 h4+ Kf2 Kf4 Bd4 Ke4 Bc3 Kf4 Bd4 Ke4 Bc3 Kf4 Bd4 Ke4 Bc3 
00:03:02.0	1,40	22	75996244	f6 h4 Kg6 exf6 Kxf6 Kg1 Kf5 hxg5 Kxg5 Bb4 Kf5 Be1 Kg4 Kh2 Be4 Bf2 h4 Be1 h3 Bd2 f3 Be3 
00:05:07.2	1,78	22	128005261	h4 Bb4 Kg6 Be7 Kf5 Bb4 g4 hxg4+ Ke4 Be7 h3 Bg5 Bc4 Kg1 f3 Bh4 Kxd4 Kh2 Ke3 e6 Bxe6 Kxh3 
00:07:36.2	1,79	23	189641372	h4 Bb4 Kg6 Be7 Kf5 Kg1 f3 Kf2 g4 Bxh4 gxh3 Bg3 Ke4 Bh2 Kxd4 Bg3 Ke4 Bh2 Be6 Kg3 Bc4 Bg1 Bd5 Kxh3 
00:11:01.1	1,79	24	274409981	h4 Bb4 Kg6 Be7 Kf5 Kg1 f3 Kf2 g4 Bxh4 gxh3 Bg3 Ke4 Bh2 Kxd4 Bg3 Ke4 Bh2 Be6 Bg1 Bg4 Bh2 
00:23:06.0	1,84	25	566343365	h4 Bb4 Kg6 Be7 Kf5 Kg1 f3 Kf2 g4 Bxh4 gxh3 Bg3 Ke4 Bh2 Kxd4 Bg3 Ke4 Bh2 Be6 Bg1 Kf4 Bh2+ Kg4 
00:23:52.3	1,84	25	585645180	h4 Bb4 Kg6 Be7 Kf5 Kg1 f3 Kf2 g4 Bxh4 gxh3 Bg3 Ke4 Bh2 Kxd4 Bg3 Ke4 Bh2 Be6 Bg1 Kf4 Bh2+ Kg4

// Christophe
Top
User avatar
smirobth
Posts: 2307
Joined: Wed Mar 08, 2006 8:41 pm
Location: Brownsville Texas USA

Re: hard endgame test position

Post by smirobth »

BBauer wrote:Again a very interesting position.
Programs have great problems playing endgames with bishops of opposite coulors. In the book of Awerbach I found an endgame where the program plays the rihgt move in 0 sec, but fails to find the right continuation. Often 2 more pawns are not enough to win, but the program is lured by a high score.

To your position:
I found

1. ...h4 2.Bb4 Kg6
3.Be7 Kf5 4.Bb4 Ke4
5.Kg1 Kf3 6.Be7 g4
7.hxg4 h3 8.Kh2 Ke2
9.Bh4 f3 10.Kxh3 Be6
11.Bg3 f2 12.Bxf2 Kxf2
13.Kh4 0-1

Do you agree?

kind regards
Bernhard
Hi Bernhard,
The first 3 moves you give for Black are definately correct. I would have to check the rest, since Black's winning plan changes slightly depending on what White does. However I suspect the remaining moves you give are probably correct as well.
- Robin Smith
Top
User avatar
smirobth
Posts: 2307
Joined: Wed Mar 08, 2006 8:41 pm
Location: Brownsville Texas USA

Re: hard endgame test position

Post by smirobth »

tiger wrote: I don't know if it's good enough to convince the experts, but here is Chess Tiger 2007.1 output on P3-933MHz (1 core), 48Mb hash tables, no endgame database:

Code: Select all

Time         Score Depth Nodes    PV
00:03:02.0	1,40	22	75996244	f6 h4 Kg6 exf6 Kxf6 Kg1 Kf5 hxg5 Kxg5 Bb4 Kf5 Be1 Kg4 Kh2 Be4 Bf2 h4 Be1 h3 Bd2 f3 Be3 
00:05:07.2	1,78	22	128005261	h4 Bb4 Kg6 Be7 Kf5 Bb4 g4 hxg4+ Ke4 Be7 h3 Bg5 Bc4 Kg1 f3 Bh4 Kxd4 Kh2 Ke3 e6 Bxe6 Kxh3
// Christophe
1...f6 only draws, but finding h4 in only 5 minutes is very good.
- Robin Smith
Top

Return to “Computer Chess Club: General Topics”