For Giraffe, not searching them is much better -
Code: Select all
Rank Name Elo + - games score oppo. draws
1 Giraffe orig 37 6 6 3174 61% -37 33%
2 Giraffe fullqs -37 6 6 3174 39% 37 33%
Moderators: hgm, Rebel, chrisw
Code: Select all
Rank Name Elo + - games score oppo. draws
1 Giraffe orig 37 6 6 3174 61% -37 33%
2 Giraffe fullqs -37 6 6 3174 39% 37 33%
Code: Select all
Rank Name Elo + - games score oppo. draws
1 Giraffe qschecks 16 6 6 3210 55% -16 32%
2 Giraffe orig -16 6 6 3210 45% 16 32%
Most people do not search SEE-losing captures in QS. It is impressive you managed to build such a strong engine without doing this, as the QS can really explode if you search all captures all the time.matthewlai wrote:Do you search SEE-losing captures in QS?
matthewlai wrote:And extending all non-SEE-losing checks in QS is an improvement.
If QS gets a position where the moving side is in check, it does a full search with node budget 1 (equivalent to depth 1) search instead.
Code: Select all
Rank Name Elo + - games score oppo. draws 1 Giraffe qschecks 16 6 6 3210 55% -16 32% 2 Giraffe orig -16 6 6 3210 45% 16 32%
Same as my other response. When you use an atypical search approach, things change. This likely suggests a flaw in the basic search idea you use with this "node budget" stuff, because not including losing captures (SEE) is a clear win in every test I have done, with each test having about 10x more games than your sample.matthewlai wrote:Do you search SEE-losing captures in QS? I have always not searched them, and just did some quick test.
For Giraffe, not searching them is much better -Code: Select all
Rank Name Elo + - games score oppo. draws 1 Giraffe orig 37 6 6 3174 61% -37 33% 2 Giraffe fullqs -37 6 6 3174 39% 37 33%
That is certainly possible.bob wrote:I'd bet this is unique to your atypical search approach, rather than being of use to everyone.
My test showed that NOT searching losing captures is a clear win. I think you read the results wrong.bob wrote: Same as my other response. When you use an atypical search approach, things change. This likely suggests a flaw in the basic search idea you use with this "node budget" stuff, because not including losing captures (SEE) is a clear win in every test I have done, with each test having about 10x more games than your sample.
Ah no I have always not searched losing captures. I just decided to do some tests to verify that's indeed a good idea (it was).BubbaTough wrote: Most people do not search SEE-losing captures in QS. It is impressive you managed to build such a strong engine without doing this, as the QS can really explode if you search all captures all the time.
-Sam
You are right. I mis-read. This is such an old and well-known idea it looked like you were doing that and had changed...matthewlai wrote:My test showed that NOT searching losing captures is a clear win. I think you read the results wrong.bob wrote: Same as my other response. When you use an atypical search approach, things change. This likely suggests a flaw in the basic search idea you use with this "node budget" stuff, because not including losing captures (SEE) is a clear win in every test I have done, with each test having about 10x more games than your sample.
Perhaps you searched for a way to increase Nodes / second for Giraffe. But just use minimax instead of alphabeta. Argument: it improves move ordering.matthewlai wrote:Ah no I have always not searched losing captures. I just decided to do some tests to verify that's indeed a good idea (it was).BubbaTough wrote: Most people do not search SEE-losing captures in QS. It is impressive you managed to build such a strong engine without doing this, as the QS can really explode if you search all captures all the time.
-Sam
I made this post because a while ago someone posted something about searching all captures (I can't find that post anymore).