Well I still don't see much use in storing them because they aren't "real" game positions. Though of course if the side to move is handled correctly, they won't cause harm. Maybe it's also a question of how much memory you have to spare even for unreal game positions.Evert wrote:It's a little disingenious to say that this is what you suggested though, since what you said was "positions that arise from a null-move don't. Belong in the main transposition table", which is wrong.
how can IID with null allowed cause wrong mate move from tt
Moderators: hgm, Rebel, chrisw
-
- Posts: 2488
- Joined: Tue Aug 30, 2016 8:19 pm
- Full name: Rasmus Althoff
Re: how can IID with null allowed cause wrong mate move from
-
- Posts: 2488
- Joined: Tue Aug 30, 2016 8:19 pm
- Full name: Rasmus Althoff
Re: how can IID with null allowed cause wrong mate move from
I've measured that for the 3 ply depth pre-search, which is plain alpha-beta Negamax with move sorting. During normal openings, the number of QS entry nodes went down by 40%.Evert wrote:How big is this effect? I would expect killers and hash moves to dominate very quickly.
Doesn't sound much, given depth==3, but I have seen positions where the pre-sorting alone took more than 2 seconds. Which I noticed because I hadn't expected this, hadn't been triggering the 2 seconds watchdog during pre-search and sometimes got unexpected watchdog system resets. In blitz games, 2 seconds for the sorting alone is a lot.
-
- Posts: 2929
- Joined: Sat Jan 22, 2011 12:42 am
- Location: NL
Re: how can IID with null allowed cause wrong mate move from
What does that have to do with anything? Null move isn't legal, but that doesn't stop anyone from trying it.Ras wrote: Well I still don't see much use in storing them because they aren't "real" game positions.
The point is that the positions following null-move are still searched normally. As such, anything that helps the search normally will help there.
-
- Posts: 2488
- Joined: Tue Aug 30, 2016 8:19 pm
- Full name: Rasmus Althoff
Re: how can IID with null allowed cause wrong mate move from
The issue is that IMO, it doesn't make sense to do a deep search with positions that are illegal anyway. But if the search isn't deep, then there isn't much to transpose, so hash tables won't be that useful (and entries will be blocked).Evert wrote:Null move isn't legal, but that doesn't stop anyone from trying it.
Well if the null search is deep enough, yes. But I'd invest the limited time rather in deeper search for positions that actually could arise on the board.As such, anything that helps the search normally will help there.
-
- Posts: 373
- Joined: Wed Mar 22, 2006 10:17 am
- Location: Novi Sad, Serbia
- Full name: Karlo Balla
Re: how can IID with null allowed cause wrong mate move from
It seems that you are trying to disprove the well proven conceptRas wrote:The issue is that IMO, it doesn't make sense to do a deep search with positions that are illegal anyway. But if the search isn't deep, then there isn't much to transpose, so hash tables won't be that useful (and entries will be blocked).Evert wrote:Null move isn't legal, but that doesn't stop anyone from trying it.
Well if the null search is deep enough, yes. But I'd invest the limited time rather in deeper search for positions that actually could arise on the board.As such, anything that helps the search normally will help there.
The problem is that no one knows in advance which positions are relevant.
Best Regards,
Karlo Balla Jr.
Karlo Balla Jr.