Where can I download the sources of Crafty with SSE2 suppo..

Archived from groups: rec.games.chess.computer (More info?)

SUBJ.

Where Can I get the new Crafty sources with SSE2 and Multi-CPU Support.

Archived from groups: rec.games.chess.computer (More info?)

Thanks.

I've found the macro's that enable multi-threading, but no SSE-code.
Do you mean, that Crafty has no SSE2-code at all? And SSE2-Version was
simply build with compiler's optimization for SSE2?

Archived from groups: rec.games.chess.computer (More info?)

> I can't find a reason how Crafty could get advantage from using SSE2
> instructions. This is chess engine, not a multimedia tool.
Why not? SSE2 can work with integers, has logic and shift operations.
I think it could be used to optimize some bitboard operations.

> I believe, sse2 means only the target CPU optimization...
OK. I suppose, you are right.

Archived from groups: rec.games.chess.computer (More info?)

<Akorps@aol.com> wrote in message
news:1114698177.446759.284330@f14g2000cwb.googlegroups.com...
> Speaking of Crafty, another possible use of null moves is as a measure
> of how tactical or strategic a position is, and what kinds of moves one
> might need to search for. The less the value of the position changes
> after a null move (excluding zugzwang positions), the more likely the
> position is to be locked up or of a strategic character. The more the

You do not get that information back from a nullmove when using a window
[beta-1;beta]

All you get back is a fail high or a fail low. No other information. You
don't have at all garantuee even that the 'thread' in the position is giving
the cutoff. If just the static evaluation is >= beta for the opponent all he
needs to do to refute your nullmove is nullmove himself.

> value of the position changes (indicating the value of a tempo is
> greater) the more likely sacrifices are to be correct, and tactical
> lines might need to be searched more deeply. Maybe Hyatt can enlighten
> us on these topics.
>
> I wonder how Shredder uses null moves, anyone know?

Not really, but you might consider using double nullmove.