Author: Tord Romstad
Date: 04:37:48 04/21/04
Go up one level in this thread
On April 21, 2004 at 07:14:55, Sune Fischer wrote: >On April 21, 2004 at 05:44:51, Tord Romstad wrote: > >>On April 20, 2004 at 19:57:00, Sune Fischer wrote: >> >>>>I used to do static mate detection in the past, but had to sacrifice it when >>>>I simplified my attack tables. >>> >>>As always it's a trade off, your new tables are probably faster? :) >> >>Yes, though not very much. The effect on playing strength seemed to be more >>or less neutral. When there is no measurable difference in strength between >>two different ways to do something, I tend to prefer the simplest solution, >>in order to reduce the number of potential bugs. > >I figured the easiest of all was to get rid of them entirely. :) And I am trying to see how far I can get without them in my new engine. >>>>This is strange. I have very rarely seen something like this. Do you have >>>>any examples of positions where this occurs? How much did you extend for >>>>mate threats? >>> >>>It happens pretty much in all attacking positions, eg. wac141 takes a lot longer >>>to solve with null-threat extension on. >>>I use half a ply for the extension. >> >>Have you tried to make this more dynamic, like I do? I haven't tested this >>very thoroughly, but intuitively it seems plausible that it is more useful >>to extend for mate threats when the side to move has an advantage. It might >>be a good idea to let the amount of extension depend on the static eval. > >Yes I have tried this. The result on wac141 is that it still finds the right >move fast, but efter that it has a high score and takes forever to find the >mate. Yes, I am familiar with this effect. Despite its low NPS count, Gothmog is faster than most amateur engines at finding the right moves in tactical test suites. But it is usually extremely slow at announcing mates. For instance, in WAC141, Gothmog 0.4.8d needs 13,380 nodes and 0.15 seconds to find the right move, but 1,375,427 nodes and 7.19 seconds to announce mate in at most 63 moves, and 2,648,487 nodes and 13.99 seconds to push this down to mate in 6. >I'm not convinced that the overall effect is a good one, once you really have a >hold of your opponent it is better to kill him off sooner rather than later. >(see also my post to Uri) I've never seen any bad effects of it, except for the cosmetical problem of slow mate announcements. As always, YMMV. >>Very strange. Here are my results for WAC141 with mate threat extension >>on and off: >> >>Mate extension on, BM extension on: Solved in 6 plies, 13380 nodes. >>Mate extension on, BM extension off: Solved in 7 plies, 48236 nodes. >>Mate extension off, BM extension on: Solved in 9 plies, 156929 nodes. >>Mate extension off, BM extension off: Solved in 9 plies, 156937 nodes. >> >>As you can see, the combination of mate extensions and BM extensions helps >>Gothmog find Qxf4 3 plies earlier, and in less than 10% of the nodes. This >>is by no means unusual in such positions. Mate extensions improve the speed >>tremendously in positions where checkmates are important, and seems to have >>very little cost in other positions. > >Strange, I can't reproduce this effect at all. >I've read of others who didn't get anything from this so I though it was normal. > >>I had very strange results with the BM extension. I found very few test >>positions where Gothmog performed better with the extension enabled, but >>when playing games the version with the extension consistently scored >>slightly better. > >Oh that is interesting, I don't think I ever gave it a chance in real play >because it didn't seem to work even on position where I thought it should work. > >>I haven't noticed this problem very often. It probably depends on your move >>ordering. In the node following a null move, if the null move failed low >>by a big margin two plies earlier, I search the last refutation move directly >>after the hash table move. > >Yeah it could be related to move ordering, although the extension is used in >Crafty and Crafty doesn't have this kind of move ordering. Are you sure it is used by Crafty? I just had a brief look in search.c in Crafty 19.12, and couldn't see anything resembling BM extensions. >I also tried searching the refutation move first, but it didn't have a very high >cutoff percentage. I feared that searching it first would reduce general move >ordering and searching it later would trigger a lot of needless threat >extensions on the first moves. I actually found the opposite -- it seemed to *improve* my general move ordering. Perhaps this just proves that my general move ordering is really bad. Tord
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.