Rybka endgame behaviour

[Marek Soszynski]

A couple of Rybka oddities...

[D]8/8/4k2p/1ppRn3/2P4P/1P6/8/2K5 w - - 0 1

On my system, Rybka takes surprisingly long to avoid 1.Rxe5+??

---

[D]8/6K1/7P/6k1/8/8/8/8 w - - 0 0

I suspect this kind of thing is old news to many of you, but Rybka (with tablebases) says mate in 9, which is incorrect.

[peter]

Really silly!
Deep fritz 11 with it's own endgame turbo almost at once sees 1.Rxe5 leading to 0.00, playing 1.Kd1, which should win for example like this:

8/8/4k2p/1ppRn3/2P4P/1P6/8/2K5 w - - 0 1

1. Kd1 bxc4 2. bxc4 Nxc4 3. Rxc5 Ne5 4. Ke2 Ng6 5. h5 Ne5 6. Rb5 Kf5 7. Ke3 Ke6 8. Kf4 Nd7 9. Kf3 Nf6 10. Kg3 Ne4+ 11. Kh4 Nd6 12. Rb6 Ke5 13. Kg4 Kd5 14. Rb8 Ke5 15. Rh8

R3 with the same tablebases (fritz endgame turbo, complete 5, incomplete 6MOB) and option preserve hash and use tbs frequently beeing shown the remis- 4MOB and going back step by step:

8/8/4k2p/1ppRn3/2P4P/1P6/8/2K5 w - - 0 1

Analysis by Rybka 3 32-bit A:

1.Rxe5+ Kxe5 2.cxb5 Kd6 3.Kc2 Kc7 4.b4 Kb6 5.bxc5+
+- (2.66) Tiefe: 6 00:00:00
+- (3.15) Tiefe: 24 00:00:17 238kN, tb=1442

Rybka, you've goofed it again

[ernest]

Analysis by Rybka 3 32-bit A:

1.Rxe5+ Kxe5 2.cxb5 Kd6 3.Kc2 Kc7 4.b4 Kb6 5.bxc5+
+- (2.66) Tiefe: 6 00:00:00
+- (3.15) Tiefe: 24 00:00:17 238kN, tb=1442

Rybka, you've goofed it again

Well... give it a little more time:
8/8/4k2p/1ppRn3/2P4P/1P6/8/2K5 w - - 0 1
Analysis by Rybka 3 32-bit:
..............
1.Rxe5+ Kxe5[] 2.cxb5[] Kd6 3.Kd2 Kc7 4.Kc3 Kb7 5.Kc4 Kb6[] 6.b4[] cxb4[] 7.Kxb4 h5 8.Kc4 Kc7 9.Kc5 Kb7[] 10.b6 Kb8 11.Kc6 Kc8 12.Kb5 Kb8 13.Kb4 Kb7
+- (2.86) Depth: 24 00:00:25 2554kN, tb=2783
1.Rxe5+ Kxe5[] 2.cxb5[] Kd6 3.Kd2 Kc7 4.Kc3 Kb7 5.Kc4 Kb6[] 6.b4[] cxb4[] 7.Kxb4 h5 8.Kc4 Kc7 9.Kc5 Kb7[] 10.b6 Kb8 11.Kc6 Kc8 12.Kb5 Kb8 13.Kb4 Kb7
+- (2.86) Depth: 25 00:00:43 4352kN, tb=4518
1.Rxe5+ Kxe5[] 2.cxb5[] Kd6 3.Kd2 Kc7 4.Kc3 Kb7 5.Kc4 Kb6[] 6.b4[] cxb4[] 7.Kxb4 h5 8.Kc4 Kc7 9.Kc5 Kb7[] 10.b6 Kb8 11.Kc6 Kc8 12.Kb5 Kb8 13.Kb4 Kb7
+- (2.86) Depth: 26 00:01:02 6186kN, tb=7735
1.Rxe5+ Kxe5[] 2.cxb5[] Kd6 3.Kd2 Kc7 4.Kc3 Kb6 5.Kc4
+- (2.86) Depth: 27 00:01:35 9376kN, tb=12833
1.Kd1
+- (3.06 !) Depth: 27 00:04:32 26541kN, tb=79691
1.Kd1
+- (3.26 !) Depth: 27 00:06:28 35104kN, tb=177628
1.Kd1
+- (3.66 !) Depth: 27 00:09:46 49326kN, tb=317838
1.Kd1
+- (4.46 !) Depth: 27 00:17:01 83516kN, tb=550529
1.Kd1 bxc4 2.bxc4[] Nxc4[] 3.Rh5[] Ne3+ 4.Ke2[] Nf5 5.Kf3[] c4 6.Kf4[] Ne7 7.Rxh6+[] Kd5 8.Rh7[] Ke6 9.h5 c3 10.Rg7 Nd5+ 11.Kf3 Nf6 12.h6 c2 13.Rc7 Kf5 14.Rxc2 Kg6 15.Rc6 Kg5 16.Ra6
+- (4.46) Depth: 27 00:22:43 131mN, tb=642408
1.Kd1
+- (4.66 !) Depth: 28 00:35:59 206mN, tb=1032325

[peter]

Well... give it a little more time:

Hi Ernest!
I guess the time more you gave it more was also with solution in hash?
With empty hash (1024, not 1536, you remember, thanks again for that ) on quad:

8/8/4k2p/1ppRn3/2P4P/1P6/8/2K5 w - - 0 1

Analysis by Rybka 3 32-bit:

1.Kc2 bxc4 2.bxc4 Sxc4 3.Txc5 Se5 4.Kc3 h5 5.Tb5 Kf6
+- (1.89) Tiefe: 6 00:00:00 2kN
+- (1.89) Tiefe: 8 00:00:00 6kN
1.Kd1 bxc4 2.bxc4[] Sxc4 3.Txc5[] Se5 4.Ke2 Sg6 5.h5 Se5 6.Ke3 Sg4+ 7.Kf4 Sf6 8.Te5+ Kd6 9.Ta5 Ke6 10.Kf3
+- (2.09 !) Tiefe: 8 00:00:00 8kN
+- (2.18) Tiefe: 13 00:00:00 75kN, tb=31
1.Txe5+ Kxe5[] 2.cxb5[] Kd6 3.Kd2 Kc7 4.Kd3 Kb7 5.Kc4 Kb6[] 6.b4[] cxb4[] 7.Kxb4 h5[] 8.Kc4 Kb7 9.Kc5 Kc7
+- (2.44 !) Tiefe: 13 00:00:00 87kN, tb=37
+- (3.14) Tiefe: 30 00:03:47 23046kN, tb=33961

Have to admit I saw it quicker once also, but in this my second try it took 9 min. I guess on MP it's a typical matter of pure accident, when R3 consideres Rx winning and when in this case even starts calculating alternatives again.
Best time I got with 2variants-mode, in this case R3 saw Rx beeing 0.00 in 30 sec.
Have a nice New Year's Eve, Ernest!

[ernest]

I guess the time more you gave it more was also with solution in hash?

No, of course not, Peter. I always clear hash before.

But this was my first and only try with my Core 2 Duo (2 cores used) @ 3GHz (512 MB hash)
Maybe I just was lucky...

[beachknight]

Code: Select all

38: 8/8/4k2p/1ppRn3/2P4P/1P6/8/2K - Deep Fritz 11
8/8/4k2p/1ppRn3/2P4P/1P6/8/2K5 w - - 0 1

Analysis by Deep Fritz 11:

1.Kd1 bxc4 2.bxc4 Nxc4 3.Rxc5 Ne5 4.Ke2 Ng6 5.h5 Ne5 6.Ke3 Nd7 7.Ra5 Nf6 8.Kf3 Kd6 9.Kg3 Ke6 10.Kh4 Ne4 11.Kg4 Nf6+ 12.Kf3 
  +-  (2.48)   Depth: 28/37   00:02:37  62260kN, tb=205433
1.Kd1 bxc4 2.bxc4 Nxc4 3.Rxc5 Ne5 4.Ke2 Ng6 5.h5 Ne5 6.Ke3 Nd7 7.Ra5 Nf6 8.Kf3 Kd6 9.Kg3 Ke6 10.Kh4 Ne4 11.Kg4 Nf6+ 12.Kf3 
  +-  (2.48)   Depth: 29/41   00:03:02  86796kN, tb=264173
1.Kd1 bxc4 2.bxc4 Nxc4 3.Rxc5 Ne5 4.Ke2 Ng6 5.h5 Ne5 6.Ke3 Nd7 7.Ra5 Nf6 8.Kf3 Kd6 9.Kg3 Ke6 10.Kh4 Ne8 11.Kg4 Nf6+ 12.Kf3 
  +-  (2.47)   Depth: 30/42   00:03:45  134mN, tb=377643
1.Kd1 bxc4 2.bxc4 Nxc4 3.Rxc5 Ne5 4.Ke2 Ng6 5.h5 Ne5 6.Ke3 Nd7 7.Ra5 Nf6 8.Kf3 Kd6 9.Kg3 Ke6 10.Kf3 
  +-  (2.48)   Depth: 31/43   00:05:28  243mN, tb=655082
1.Kd1 bxc4 2.bxc4 Nxc4 3.Rh5 Ne3+ 4.Ke2 Ng4 5.Kf3 Nf6 6.Rxh6 Kf5 7.Rh8 Ke5 8.Rc8 Kd5 9.Ra8 Ke5 
  +-  (3.52)   Depth: 32/55   00:09:47  705mN, tb=2647349
1.Kd1 bxc4 2.bxc4 Nxc4 3.Rh5 Kf7 4.Rxc5 Nd6 5.Ke2 Kf6 6.Kf3 Nf5 7.h5 Nd6 8.Kg4 Kf7 9.Ra5 Kf6 10.Kf4 Ke6 11.Re5+ Kf6 12.Rc5 Ke6 13.Kg4 Kf6 14.Ra5 Ke6 15.Kf4 Ne8 
  +-  (3.52)   Depth: 33/52   00:12:17  1303mN, tb=3818326
1.Kd1 bxc4 2.bxc4 Nxc4 3.Rh5 Kf7 4.Rxc5 Nd6 5.Ke2 Kf6 6.Kf3 Kg7 7.Kf4 Kf6 8.Rc6 Ke6 9.h5 Kd5 10.Rb6 Ke6 11.Ra6 Kd5 12.Ra7 Ke6 13.Rg7 Nf7 14.Rg6+ Ke7 15.Ra6 Nd6 
  +-  (3.55)   Depth: 34/51   00:15:35  2332mN, tb=5620041

(, chessengines20 28.12.2008)

[beachknight]

1.Kd1 bxc4 2.bxc4 Nxc4 3.Rh5 Kf7 4.Rxc5 Nd6 5.Ke2 Kf6 6.Kf3 Kg7 7.Kf4 Kf6 8.Rc6 Ke6 9.h5 Kd7 10.Rb6 Ke7 11.Ra6 Ke6 12.Ra7 Kf6 13.Ra8 Kf7 14.Ra6 Ke6
+- (4.21) Depth: 35/53 00:23:47 4672mN, tb=10836752

[beachknight]

Code: Select all

38: 8/8/4k2p/1ppRn3/2P4P/1P6/8/2K - Deep Fritz 11
8/8/4k2p/1ppRn3/2P4P/1P6/8/2K5 w - - 0 1

Analysis by Rybka 3:

1.Rxe5+ Kxe5[] 2.cxb5[] Kd6 3.Kc2 Kc7 4.Kd3 Kb7 5.Kc3 Kc7 6.Kd3 Kb7 7.Kc3 Kc7 8.Kd3 Kb7 9.Kc3 Kc7 10.Kd3 Kb7 11.Kc3 Kc7 12.Kd3 Kb7 13.Kc3 Kc7 14.Kd3 Kb7 15.Kc3 Kc7 16.Kd3 
  +-  (2.48)   Depth: 20   00:03:23  29kN, tb=6
1.Kd1 
  +-  (2.68 !)   Depth: 20   00:04:35  109kN, tb=102
1.Kd1 
  +-  (2.88 !)   Depth: 20   00:04:54  281kN, tb=1019
1.Kd1 
  +-  (3.08)   Depth: 20   00:05:02  504kN, tb=1611
1.Kd1 
  +-  (3.08)   Depth: 21   00:05:04  610kN, tb=1665
1.Kd1 
  +-  (3.08)   Depth: 22   00:05:05  681kN, tb=1689
1.Kd1 
  +-  (3.08)   Depth: 23   00:05:05  757kN, tb=1716
1.Kd1 
  +-  (3.08)   Depth: 24   00:05:05  757kN, tb=1716
1.Kd1 
  +-  (3.28 !)   Depth: 25   00:06:07  4074kN, tb=19068
1.Kd1 
  +-  (3.48 !)   Depth: 25   00:06:47  7311kN, tb=34066
1.Kd1 bxc4 2.bxc4[] Nxc4[] 3.Rh5 Ne3+ 4.Ke2[] Nf5 5.Kf3[] c4 6.Kf4[] Ne7 7.Rxh6+[] Kd5 8.Rh7[] 
  +-  (3.76)   Depth: 25   00:07:44  13314kN, tb=64132
1.Kd1 
  +-  (3.96 !)   Depth: 26   00:09:04  23438kN, tb=104478
1.Kd1 bxc4 2.bxc4[] Nxc4 3.Rh5[] Ne3+ 4.Ke2[] Nf5 5.Kf3[] c4 6.Kf4[] Ne7 7.Rxh6+[] Kd5 8.Rh7[] Ke6 9.Ke4[] c3 10.Kd3[] Kf6 11.Rh6+[] Kf5 12.Rh5+ Kg4 13.Re5[] Nc6 14.Re4+ Kh5 15.Kxc3 Nd8 16.Kd4 
  +-  (4.04)   Depth: 26   00:10:27  33039kN, tb=150036
1.Kd1 bxc4 2.bxc4[] Nxc4 3.Rh5[] Ne3+ 4.Ke2[] Nf5 5.Kf3[] c4 6.Kf4[] Ne7 7.Rxh6+[] Kd5 8.Rh7[] Ke6 9.Ke4[] c3 10.Kd3[] Kf6 11.Rh6+[] Kf5 12.Rh5+ Kg4 13.Re5[] Nc6 14.Re4+ Kh5 15.Kxc3 Nd8 16.Kd4 
  +-  (4.04)   Depth: 27   00:12:07  47232kN, tb=173608

(, chessengines20 28.12.2008)

tb = frequently
x64

[beachknight]

1.Kd1
+- (4.24 !) Depth: 28 00:17:02 87343kN, tb=291158
1.Kd1 bxc4 2.bxc4[] Nxc4[] 3.Rh5[] Ne3+ 4.Ke2[] Ng4 5.Rxc5 Ne5 6.Ke3 Nf7 7.Ke4
+- (4.24) Depth: 28 00:19:24 106mN, tb=363524