Author: Odd Gunnar Malin
Date: 05:57:11 02/25/02
Go up one level in this thread
On February 25, 2002 at 08:52:59, Uri Blass wrote:
>On February 25, 2002 at 08:50:27, Odd Gunnar Malin wrote:
>
>>On February 25, 2002 at 08:20:25, Benny Antonsson wrote:
>>
>>>Hi !
>>>
>>>I have a question about how to count the nodes when using alphaBeta and
>>>quiscense searching. Here is what we do (pseudo code):
>>>
>>>function alphaBeta()
>>>{
>>> nodeCount++;
>>>
>>> value = quisce();
>>>}
>>>
>>>function quisce()
>>>{
>>> nodeCount++;
>>>}
>>>
>>>Wouldn't this result in counting the first quisce-node twice ??
>>
>>Yes.
>>I count qnodes after I do the move.
>>
>>GenerateCapturemoves(...);
>>while (movesreminding)
>>{
>> if (doLegalMove(..))
>> {
>> qnodes++;
>> score=-qsearch(..);
>> undomove(..);
>> ...
>> }
>>}
>>
>>I count the normal nodes like you do.
>>
>>It was a discussion here some time ago that some was counting the move they
>>generate regardless if this was pruned away or not. I think this would be
>>inaccurate and only for better showup in the nps. Instead I have a movegenerator
>>test for timing the move/undomove. This is also usefull for testing correctness
>>of the movegenerator.
>>
>>Odd Gunnar
>
>I have no moves that are pruned away.
>I make only legal moves.
>
>Uri
I thought it was you but wasn't sure :)
OGM
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.