Author: Jonas Cohonas
Date: 18:32:21 12/03/01
The sad thing is that when you get a speedup with DF that is acceptable, then there is no guarantee that you can reproduce it with the same position, after turning off you comp and starting it again, it might even be much worse than the speed you get with one processor. I suspect that it has something to do with it's way of handling hash, since if you turn on infinite analysis and get a speedup worth talking about, then turning off infinite and start again, the speedup has gone. If you try clearing hash before you restart infinite analysis you will get the speedup again.... By running DF in the F7 GUI i managed to get a reasonable speedup, but the result is sad when you see that it took longer for DF on 2 processors to solve this position: (1processor) BS2830.05 2r1r3/p3bk1p/1pnqpppB/3n4/3P2Q1/PB3N2/1P3PPP/3RR1K1 w - - 0 1 Analysis by Deep Fritz: 1.Bxd5 Qxd5 2.Rc1 Bd6 ² (0.41) Depth: 3/10 00:00:00 1.Rc1 b5 2.Bxd5 exd5 ² (0.53) Depth: 3/15 00:00:00 1kN 1.Rc1 b5 2.Bxd5 exd5 ² (0.53) Depth: 3/15 00:00:00 1kN 1.Rc1 b5 2.Bxd5 exd5 ² (0.53) Depth: 3/15 00:00:00 1kN 1.Rc1 b5 2.Bxd5 exd5 ² (0.53) Depth: 3/15 00:00:00 1kN 1.Rc1 b5 2.Bxd5 Qxd5 3.Bf4 ² (0.50) Depth: 4/15 00:00:00 2kN 1.Rc1 b5 2.Bxd5 Qxd5 3.Bf4 ² (0.50) Depth: 4/15 00:00:00 2kN 1.Rc1 b5 2.Bxd5 Qxd5 3.Bf4 ² (0.50) Depth: 4/15 00:00:00 2kN 1.Rc1 b5 2.Rc5 a6 3.Bxd5 exd5 ² (0.47) Depth: 5/18 00:00:00 9kN 1.Bc4! ² (0.50) Depth: 5/18 00:00:00 13kN 1.Bc4 Bf8 2.Bxf8 Kxf8 3.Bxd5 Qxd5 ² (0.56) Depth: 5/19 00:00:00 18kN 1.Bc4 Bf8 2.Bd2 Kg8 3.Rc1 Bg7 ² (0.53) Depth: 6/19 00:00:00 48kN 1.Bc4 Bf8 2.Bd2 Kg8 3.Rc1 Bg7 4.b3 ² (0.53) Depth: 7/23 00:00:00 183kN 1.Bc4 Bf8 2.Bd2 Kg8 3.Rc1 Bg7 4.b3 a5 ² (0.56) Depth: 8/25 00:00:00 815kN 1.Bc4 Bf8 2.Bd2 Kg8 3.Rc1 Bg7 4.Qe4 Re7 5.b3 ² (0.53) Depth: 9/27 00:00:01 1506kN 1.Qh3! ² (0.56) Depth: 9/28 00:00:02 2977kN 1.Qh3 b5 2.Rxe6 Qxe6 3.Re1 Qf5 4.Qxf5 gxf5 5.Bxd5+ Kg6 6.Bf4 Kg7 ± (0.78) Depth: 10/31 00:00:13 13495kN 1.Qh3 Rh8 2.Bc4 Na5 3.Ba6 Rb8 4.a4 Nc6 5.Rc1 Rhd8 6.Bb5 ± (0.75) Depth: 11/32 00:00:23 24160kN 1.Qh3 Rh8 2.Nd2 Qd7 3.Nb1 f5 4.Nc3 Rhd8 5.Nxd5 exd5 6.Qf3 ± (0.88) Depth: 12/33 00:00:49 51688kN 1.Qh3 Rh8 2.Nd2 Qd7 3.Ne4 f5 4.Bxd5 exd5 5.Ng5+ Bxg5 6.Bxg5 Rce8 ± (0.88) Depth: 13/35 00:01:29 94483kN 1.Rxe6! ± (0.91) Depth: 13/35 00:02:42 177833kN 1.Rxe6 Qxe6 2.Re1 f5 3.Rxe6 fxg4 4.Bxd5 Red8 5.Rxc6+ Rxd5 6.Rxc8 gxf3 ± (1.06) Depth: 13/35 00:02:46 182921kN (, Dual MP 1600+ 04.12.2001) (2processors) BS2830.05 2r1r3/p3bk1p/1pnqpppB/3n4/3P2Q1/PB3N2/1P3PPP/3RR1K1 w - - 0 1 Analysis by Deep Fritz: 1.Bxd5! ² (0.50) Depth: 2/12 00:00:00 1.Bxd5 Qxd5 2.Rc1 Bd6 ² (0.41) Depth: 3/10 00:00:00 1.Rc1 b5 2.Bxd5 exd5 ² (0.53) Depth: 3/15 00:00:00 1kN 1.Rc1 b5 2.Bxd5 exd5 ² (0.53) Depth: 3/15 00:00:00 1kN 1.Rc1 b5 2.Bxd5 exd5 ² (0.53) Depth: 3/15 00:00:00 1kN 1.Rc1 b5 2.Bxd5 exd5 ² (0.53) Depth: 3/15 00:00:00 1kN 1.Rc1 b5 2.Bxd5 exd5 ² (0.53) Depth: 3/15 00:00:00 1kN 1.Rc1 b5 2.Bxd5 Qxd5 3.Bf4 ² (0.50) Depth: 4/15 00:00:00 2kN 1.Rc1 b5 2.Rc5 a6 3.Bxd5 exd5 ² (0.47) Depth: 5/18 00:00:00 9kN 1.Bc4! ² (0.50) Depth: 5/18 00:00:00 13kN 1.Bc4 Bf8 2.Bxf8 Kxf8 3.Bxd5 Qxd5 ² (0.56) Depth: 5/19 00:00:00 18kN 1.Bc4 Bf8 2.Bd2 Kg8 3.Rc1 Bg7 ² (0.53) Depth: 6/19 00:00:00 48kN 1.Bc4 Bf8 2.Bd2 Kg8 3.Rc1 Bg7 4.b3 ² (0.53) Depth: 7/22 00:00:00 189kN 1.Bc4 Bf8 2.Bd2 Kg8 3.Rc1 Bg7 4.b3 a5 ² (0.56) Depth: 8/25 00:00:00 659kN 1.Bc4 Bf8 2.Bd2 Kg8 3.Rc1 Bg7 4.Qe4 Re7 5.b3 ² (0.53) Depth: 9/27 00:00:01 1416kN 1.Qh3! ² (0.56) Depth: 9/30 00:00:02 2958kN 1.Qh3 Nd8 2.Nd2 Qd7 3.Nf1 Kg8 4.Ne3 Nxe3 5.Qxe3 Bd6 6.Bd5 f5 ± (0.72) Depth: 10/31 00:00:07 11383kN 1.Qh3 Nd8 2.Nd2 Qd7 3.Nf1 Qb5 4.a4 Qd7 5.Ne3 Rc7 6.Nxd5 ± (0.72) Depth: 11/29 00:00:11 16751kN 1.Qh3 Nd8 2.Nd2 Qd7 3.Nf1 Qb5 4.a4 Qd7 5.Ne3 f5 6.Rc1 Rxc1 ± (0.78) Depth: 12/34 00:00:22 34772kN 1.Qh3 Rh8 2.Nd2 Qd7 3.Ne4 f5 4.Bxd5 exd5 5.Ng5+ Bxg5 6.Bxg5 b5 ± (0.88) Depth: 13/38 00:02:27 224885kN 1.Rxe6! ± (0.91) Depth: 13/38 00:03:02 282956kN (, Dual MP 1600+ 04.12.2001) Regards Jonas
This page took 0.01 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.