What program first made decisions by node-type?
Moderators: hgm, Rebel, chrisw
-
- Posts: 5960
- Joined: Sun Jan 10, 2010 6:15 am
- Location: Maryland USA
What program first made decisions by node-type?
Which is the first program known to make distinctions in its search algorithms between PV nodes and non-PV nodes? This could be different extensions or different reductions based on node type. I don't recall hearing of this prior to the 2000s, but I could well be wrong.
-
- Posts: 4185
- Joined: Tue Mar 14, 2006 11:34 am
- Location: Ethiopia
Re: What program first made decisions by node-type?
AFAIK Fruit. I believe Fabian did it to get a stable search and PV.
No null_move at PV, no transposition cut off at PV, no futility at PV, no lmr at pv, more extensions at pv..
In general no selectivity at pv except for the extensions. Most of them have some elo cost associated with them it seems (however tiny).
Almost everybody does LMR at pv now, while I did it from the very beginning..
No null_move at PV, no transposition cut off at PV, no futility at PV, no lmr at pv, more extensions at pv..
In general no selectivity at pv except for the extensions. Most of them have some elo cost associated with them it seems (however tiny).
Almost everybody does LMR at pv now, while I did it from the very beginning..
-
- Posts: 6401
- Joined: Thu Mar 09, 2006 8:30 pm
- Location: Chicago, Illinois, USA
Re: What program first made decisions by node-type?
This is much older. Averno, according to his author, had different search functions for PV and non-PV circa 2001 and I am sure there were precedents for it.Daniel Shawul wrote:AFAIK Fruit. I believe Fabian did it to get a stable search and PV.
No null_move at PV, no transposition cut off at PV, no futility at PV, no lmr at pv, more extensions at pv..
In general no selectivity at pv except for the extensions. Most of them have some elo cost associated with them it seems (however tiny).
Almost everybody does LMR at pv now, while I did it from the very beginning..
Miguel
-
- Posts: 4185
- Joined: Tue Mar 14, 2006 11:34 am
- Location: Ethiopia
Re: What program first made decisions by node-type?
If parallel search counts , then it is way back. PV-split comes to mind.
I am sure Knuth's PV,CUT and ALL node classification have been used in many programs in different ways.
But none in a way Fruit did it to take care of its PV. I believe this is more of a style than substance. But the fact that Fruit did less extensions did boost its elo a lot IMO.
The OP's question has to more specific . Also published material paper , source code or even a forum post would be good to have when discussing things of historical matter.
Daniel
I am sure Knuth's PV,CUT and ALL node classification have been used in many programs in different ways.
But none in a way Fruit did it to take care of its PV. I believe this is more of a style than substance. But the fact that Fruit did less extensions did boost its elo a lot IMO.
The OP's question has to more specific . Also published material paper , source code or even a forum post would be good to have when discussing things of historical matter.
Daniel
-
- Posts: 2949
- Joined: Mon May 05, 2008 12:16 pm
- Location: Bordeaux (France)
- Full name: Julien Marcel
Re: What program first made decisions by node-type?
Second question on computer chess history, Larry. Are you writing a book?
"The only good bug is a dead bug." (Don Dailey)
[Blog: http://tinyurl.com/predateur ] [Facebook: http://tinyurl.com/fbpredateur ] [MacEngines: http://tinyurl.com/macengines ]
[Blog: http://tinyurl.com/predateur ] [Facebook: http://tinyurl.com/fbpredateur ] [MacEngines: http://tinyurl.com/macengines ]
-
- Posts: 5960
- Joined: Sun Jan 10, 2010 6:15 am
- Location: Maryland USA
Re: What program first made decisions by node-type?
No, I'm just curious about who should get credit for these ideas. I am writing a chess book, but not on computer chess.
-
- Posts: 2250
- Joined: Wed Mar 08, 2006 8:47 pm
- Location: Hattingen, Germany
Re: What program first made decisions by node-type?
Implicit the first programs using PVS/Negascout with some forward pruning near the tips, in PVS implementations often with different PVS/NWS routines.lkaufman wrote:Which is the first program known to make distinctions in its search algorithms between PV nodes and non-PV nodes? This could be different extensions or different reductions based on node type. I don't recall hearing of this prior to the 2000s, but I could well be wrong.
Using explicit different extensions and later reductions, I don't know. Probably already early Type-B programs like Awit and Chaos, and Lang programs or even Henne/Nitsche. Levy, Broughton and Taylor did not mention node types explicitly in their 1989 SEX paper, but probability that a terminal node becomes part of the PV.