Author: Michel Langeveld
Date: 09:05:45 01/16/04
Go up one level in this thread
On January 16, 2004 at 03:20:59, Richard Pijl wrote: >I wouldn't call this a bug, and I doubt it is really an improvement. How many >positions did you test after this change? In fact you're reducing the increment >to the history counter on the deeper search. >Beta cutoffs in a deeper search should have a much bigger impact in moveordering >than shallow searches. >Another thing you can try is to use 'depth*depth' instead of 'depth' to >increment history with. >Richard. Hi Richard, I tested 2 positions. Not a big deal but better as nothing. Depth * depth improves the starting position and made the tactical shot a tiny but slower (not in time). I think keep it in. Thanks a lot Richard! STARTING //with +=depth 8* 684 1054485 -7 i1h3 d7d6 d2d4 b8c6 b1c3 c8f5 c1f4 i8h6 9* 2601 4191174 -2 i1h3 d7d6 d2d4 i8h6 b1c3 c8f5 c1f4 b8c6 h3g5 10* 9640 14861438 -7 b1c3 d7d6 d2d4 b8c6 i1h3 i8h6 c1f4 c8f5 h3g5 h6g4 //with +=depth * depth 8* 665 1041380 -7 i1h3 d7d6 d2d4 b8c6 b1c3 c8f5 c1f4 i8h6 9* 2581 4177903 -2 i1h3 d7d6 d2d4 i8h6 b1c3 c8f5 c1f4 b8c6 h3g5 10* 8568 13334621 -3 i1h3 d7d5 d2d4 i8h6 b1c3 c8f5 c1f4 b8a6 h3g5 h6g4 //with += depth + ext (=original TSCP) 8* 701 1099674 -7 i1h3 d7d6 d2d4 b8c6 b1c3 c8f5 c1f4 i8h6 9* 2340 3745971 -1 b1c3 d7d6 d2d4 i8h6 i1j3 c8g4 c1f4 h6f5 c3e4 10* 9312 14536895 -3 b1c3 d7d5 d2d4 i8h6 i1h3 c8f5 c1f4 b8a6 h3g5 h6g4 TACTICAL //with += depth 6* 37 61106 -9990 i3j3 h5h4 d1j7 i8j7 j3i4 g2i2 i4h5 c8e8 h5g5 i2i6 //with += depth * depth 6* 37 61374 -9990 i3j3 h5h4 d1j7 i8j7 j3i4 g2i2 i4h5 c8e8 h5g5 i2i6 //with += (depth + ext) (=original TSCP) 6* 37 63690 -9990 i3j3 h5h4 d1j7 i8j7 j3i4 g2i2 i4h5 c8e8 h5g5 i2i6
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.