Author: Vincent Diepeveen
Date: 19:32:11 05/10/04
Go up one level in this thread
On May 10, 2004 at 22:18:32, K. Burcham wrote: Well he tested version 2.01 I'm at version 2.05 now which is a lot better. I fixed the nps problem now. All nodes that get considered are counted now. I previously just counted actual function calls for nodes that get fully visited (not makemoves). ++ g1-f3 procnr=1 terug=29 org=[16;17] newwindow=[16;520000] 00:00 5424 2712 0 (2) 3 (0,2) 0.183 Ng1-f3 d7-d5 e2-e3 00:00 8432 4216 0 (2) 4 (0,3) 0.001 Ng1-f3 d7-d5 e2-e3 Bc8-f5 00:00 39155 20361 0 (2) 5 (0,4) 0.086 Ng1-f3 Ng8-f6 e2-e3 e7-e6 Bf1-c4 00:00 78231 44592 0 (2) 6 (0,5) 0.001 Ng1-f3 Ng8-f6 e2-e3 e7-e6 Bf1-c4 Bf8-c5 00:00 385054 315745 0 (2) 7 (0,6) 0.132 Ng1-f3 d7-d5 e2-e3 00:01 656234 728420 0 (2) 8 (0,15) 0.001 Ng1-f3 Ng8-f6 Nb1-c3 e7-e6 d2-d4 Bf8-b4 Bc1-g5 Nb8-c6 00:02 1629421 4138730 0 (2) 9 (8,147) 0.085 Ng1-f3 e7-e6 d2-d3 Ng8-f6 Nb1-c3 Bf8 -b4 Bc1-g5 h7-h6 Bg5-f4 ++ b1-c3 procnr=1 terug=86 org=[85;86] newwindow=[85;520000] 00:03 1703828 5673748 0 (2) 9 (16,188) 0.131 Nb1-c3 d7-d5 d2-d4 Bc8-f5 Bc1-f4 e7 -e6 Ng1-f3 Bf8-b4 e2-e3 00:05 1888235 9573352 0 (2) 10 (23,230) 0.002 Nb1-c3 d7-d5 d2-d4 Bc8-f5 Bc1-f4 e 7-e6 e2-e3 Bf8-b4 a2-a3 Bb4-a5 00:27 2199404 59471892 0 (2) 11 (97,736) 0.198 Nb1-c3 d7-d5 Ng1-f3 Nb8-c6 d2-d4 f7-f6 Bc1-f4 g7-g5 Bf4-g3 Bc8-g4 e2-e3 00:46 2225875 104282281 0 (2) 12 (117,888) 0.001 Nb1-c3 d7-d5 Ng1-f3 Nb8-c6 d2-d 4 Bc8-g4 Bc1-f4 e7-e6 Nc3-b5 Ra8-c8 c2-c3 Bf8-e7 ++ g1-f3 procnr=0 terug=2 org=[1;2] newwindow=[1;520000] 01:56 2180183 252988508 0 (2) 12 (150,1012) 0.005 Ng1-f3 Ng8-f6 Nb1-c3 d7-d6 e2- e4 g7-g6 d2-d4 Bc8-g4 Bf1-e2 Bf8-g7 e4-e5 Nf6-d7 08:03 2222198 1073810812 0 (2) 13 (310,1914) 0.230 Ng1-f3 Nb8-c6 d2-d4 Ng8-f6 Bc 1-f4 Nf6-h5 Bf4-g3 d7-d5 e2-e3 Bc8-f5 Bf1-b5 First few seconds XP is swapping a bit but after a minute or 8 i guess it's stabilized at 2.2 mln nps. For version 2.01 this compared to about 190k full nodes a second. All those simple chessprograms also count every node before they hit a 'return'. I remember they rumoured schach just did a single compare then added a node and hit return. Considered nodes i spend quite a lot system time at, way more than this. So they are counted now. It's fixed for good. No more remarks diep is a slow engine please. It just carries a lot of chessknowledge, that's all. > > >so the 485 is misprint, and should read 4850 kns. > >ok thanks >kburcham > >http://www.aceshardware.com/read.jsp?id=60000285
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.