Author: enrico carrisco
Date: 14:06:43 01/01/06
Go up one level in this thread
I stated here in CCC days after Rybka was released that the nps was manipulated but lacked any real convincing examples of this. I even ran the idea by Chrilly when he was disassembing Rybka, but he had no comment on it. In the following position, we see two different evaluations for Rybka 1.0 Beta 64-bit and Rybka 1.01 Preview 2 64-bit. Of course, without knowing what undocumented changes (if any) were made from 1.0 Beta to Preview 2, it's hard to deduce anything concrete. However, Vasik has posted here a few times that nothing was changed with the strength or eval (I believe... Not trying to quote anything...) Specifically, the time-to-ply figures are odd. Preview 2 appears to slow down in time to ply while Rybka 1.0 Beta seems to 'speed up' as the depth increases (from nps figures.) The second position, below, shows the differences in nps better -- where the qsearch demands a little recognition... :) Time-to-ply is very close (unlike the first position) but the nps differs substantially... Position 1 [time-to-ply oddness]: [D]4k3/1b4R1/q3p2p/2B2p2/8/1R6/1K6/8 w - - 0 88 Rybka 1.0 beta 64 bit 512mb hash A64-2.7 (CP GUI): 00:00:00.0 2.40 3 65 Rbxb7 00:00:00.0 2.41 4 435 Rbxb7 00:00:00.0 2.65 5 1210 Rbxb7 Qe2+ Kb3 00:00:00.1 5.63 6 2476 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 00:00:00.1 5.63 7 4189 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 00:00:00.1 5.63 8 8222 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qe2+ Ka5 00:00:00.1 6.50 9 16000 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qe2+ Ka5 Qd2+ 00:00:00.3 6.56 10 41344 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qe2+ Kb6 Qb2+ 00:00:00.5 7.43 11 118209 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qe2+ Kb6 Qb2+ 00:00:01.0 7.43 12 200997 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qe2+ Kb6 Qb2+ 00:00:02.0 7.43 13 477000 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qd3+ Kb6 Qd8+ 00:00:04.0 7.49 14 987862 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qd3+ Kb6 Qd8+ 00:00:08.2 7.49 15 2073381 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qd3+ Kb6 Qd8+ 00:00:17.7 7.49 16 4778724 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qd3+ Kb6 Qd8+ 00:00:49.2 8.35 17 19664761 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qd3+ Kb6 Qb1+ 00:01:24.2 8.35 18 31087304 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qd3+ Kb6 Qb1+ 00:02:51.4 8.41 19 66442682 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qd3+ Kb6 Qb1+ Rybka 1.01 Preview 2 64 bit 512mb hash A64-2.7 (CP GUI): 00:00:00.0 2.40 3 946 Rbxb7 00:00:00.0 2.41 4 2031 Rbxb7 00:00:00.0 2.65 5 5626 Rbxb7 Qe2+ Kb3 00:00:00.0 5.63 6 10221 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 00:00:00.0 5.63 7 15896 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 00:00:00.1 5.63 8 30368 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qe2+ Ka5 00:00:00.1 6.50 9 53606 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qe2+ Ka5 Qd2+ Kb6 Qb2+ Kc6 00:00:00.2 6.56 10 111628 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qe2+ Kb6 Qb2+ Kc6 Qxg7 Rxg7 00:00:00.6 7.43 11 238776 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qe2+ Kb6 Qb2+ Kc6 Qxg7 Rxg7 f4 00:00:01.0 7.43 12 407147 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qe2+ Kb6 Qb2+ Kc6 Qxg7 Rxg7 f4 00:00:02.6 7.43 13 1136628 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qd3+ Kb6 Qd8+ Ka6 Qa8+ Kb5 Qxb7+ 00:00:04.7 7.49 14 2097110 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qd3+ Kb6 Qd8+ Ka6 Qa8+ Kb5 Qxb7+ 00:00:08.3 7.49 15 3812588 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qd3+ Kb6 Qd8+ Ka6 Qa8+ Kb5 Qxb7+ 00:00:18.0 7.49 16 8337775 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qd3+ Kb6 Qd8+ Ka6 Qa8+ Kb5 Qxb7+ 00:00:59.2 7.49 17 29186799 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qb1+ Ka5 Qxb7 Rxb7 h5 Re7+ Kf8 Rxe6+ Kf7 00:01:50.9 8.35 18 55194222 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qb1+ Ka5 Qxb7 Rxb7 f4 Re7+ Kf8 Rxe6+ Kg7 00:03:30.0 8.41 19 104388523 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qb1+ Ka5 Qxb7 Rxb7 f4 Kb5 f3 Kc6 f2 Position 2 [nps oddness]: [D]r2qkb1r/1p1b1ppp/p1nppn2/6B1/3NP3/2N5/PPPQ1PPP/2KR1B1R w kq - 0 1 Rybka 1.0 beta 64 bit 512mb hash A64-2.7 (CB GUI): 9.Nxc6 ² (0.37) Depth: 3 00:00:00 9.Kb1 ² (0.39) Depth: 3 00:00:00 9.Kb1 ² (0.32) Depth: 4 00:00:00 9.Kb1 Be7 ² (0.27) Depth: 5 00:00:00 9.Kb1 h6 10.Be3 = (0.22) Depth: 6 00:00:00 24kN 9.Be2 Be7 10.Nxc6 = (0.24) Depth: 6 00:00:00 28kN 9.Be2 Be7 10.Kb1 0-0 = (0.25) Depth: 7 00:00:00 49kN 9.Be2 Be7 10.Qe3 0-0 11.Qg3 = (0.23) Depth: 8 00:00:00 93kN 9.Be2 h6 10.Be3 Be7 11.b3 0-0 = (0.16) Depth: 9 00:00:01 215kN 9.Nxc6 Bxc6 10.f3 Qa5 11.Bf4 0-0-0 ² (0.28) Depth: 9 00:00:01 259kN 9.Nxc6 Bxc6 10.f3 Qa5 11.Kb1 Be7 12.Be2 ² (0.28) Depth: 10 00:00:02 389kN 9.Nxc6 Bxc6 10.f3 Qc7 11.Kb1 Be7 12.Bf4 Rd8 = (0.21) Depth: 11 00:00:04 674kN 9.f4 Qc7 10.Be2 Be7 11.Kb1 h6 12.Bh4 0-0 = (0.23) Depth: 11 00:00:11 1658kN 9.f4 Qc7 10.Be2 Be7 11.Kb1 h6 12.Bh4 0-0 13.Nf3 = (0.24) Depth: 12 00:00:16 2429kN 9.f4 Qc7 10.Be2 Be7 11.Kb1 h6 12.Bh4 Rd8 13.Nb3 Bc8 = (0.20) Depth: 13 00:00:27 4147kN 9.f4 Qc7 10.Be2 Be7 11.Kb1 h6 12.Bh4 Rd8 13.Nb3 Bc8 = (0.19) Depth: 14 00:01:02 9484kN 9.f4 Qc7 10.Be2 Be7 11.Kb1 0-0-0 12.Nxc6 Bxc6 13.Qe3 Kb8 = (0.20) Depth: 15 00:02:28 23322kN Rybka 1.01 Preview 2 64 bit 512mb hash A64-2.7 (CB GUI): 9.Nxc6 ² (0.37) Depth: 3 00:00:00 9.Kb1 ² (0.39) Depth: 3 00:00:00 9.Kb1 ² (0.32) Depth: 4 00:00:00 9.Kb1 Be7 ² (0.27) Depth: 5 00:00:00 9.Kb1 h6 10.Be3 = (0.22) Depth: 6 00:00:00 50kN 9.Be2 Be7 10.Nxc6 = (0.24) Depth: 6 00:00:00 61kN 9.Be2 Be7 10.Kb1 0-0 = (0.25) Depth: 7 00:00:00 103kN 9.Be2 Be7 10.Qe3 0-0 11.Qg3 = (0.23) Depth: 8 00:00:00 203kN 9.Be2 h6 10.Be3 Be7 11.b3 0-0 = (0.16) Depth: 9 00:00:01 442kN 9.Nxc6 Bxc6 10.f3 Qa5 11.Bf4 0-0-0 ² (0.28) Depth: 9 00:00:01 519kN 9.Nxc6 Bxc6 10.f3 Qa5 11.Kb1 Be7 12.Be2 ² (0.28) Depth: 10 00:00:02 752kN 9.Nxc6 Bxc6 10.f3 Qc7 11.Kb1 Be7 12.Bf4 Rd8 = (0.21) Depth: 11 00:00:04 1250kN 9.f4 Qc7 10.Be2 Be7 11.Kb1 h6 12.Bh4 0-0 = (0.23) Depth: 11 00:00:11 3124kN 9.f4 Qc7 10.Be2 Be7 11.Kb1 h6 12.Bh4 0-0 13.Nf3 = (0.24) Depth: 12 00:00:16 4578kN 9.f4 Qc7 10.Be2 Be7 11.Kb1 h6 12.Bh4 Rd8 13.Nb3 Bc8 = (0.20) Depth: 13 00:00:28 7698kN 9.f4 Qc7 10.Be2 Be7 11.Kb1 h6 12.Bh4 Rd8 13.Nb3 Bc8 14.Qe3 = (0.19) Depth: 14 00:01:02 17348kN 9.f4 Qc7 10.Be2 Be7 11.Kb1 0-0-0 12.Nxc6 Bxc6 13.Qe3 Kb8 14.e5 dxe5 15.fxe5 Nd5 = (0.20) Depth: 15 00:02:29 41641kN
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.