Stockfish v2.1 now available

[tomgdrums]

I can't seem to get the 32 bit Windows compile to work with a Polyglot book. Am I alone in my struggles?

I had this problem as well. I am not sure why it isn't reading the opening book.

BUT I am VERY Thankful to the Stockfish team for the limit strength feature so the opening book bug doesn't bother me too much.

+1

[tmokonen]

I can't seem to get the 32 bit Windows compile to work with a Polyglot book. Am I alone in my struggles?

I had this problem as well. I am not sure why it isn't reading the opening book.

BUT I am VERY Thankful to the Stockfish team for the limit strength feature so the opening book bug doesn't bother me too much.

+1

Well, a new Stockfish always makes my day. Thank you, Stockfish team.

[muxecoid]

The evaluation is more stable, i heard it's due to probe TT at PV changes.

The reported NPS is much lower than 2.0.1. Is it counted differently now?

Also eval numbers are wildly different now. Did stockfish become more materialistic?

[mcostalba]

I can't seem to get the 32 bit Windows compile to work with a Polyglot book. Am I alone in my struggles?

Have you tried with the official Salvo's book ? (http://jimablett.hostwebs.com/)

Does it works ?

[tmokonen]

I can't seem to get the 32 bit Windows compile to work with a Polyglot book. Am I alone in my struggles?

Have you tried with the official Salvo's book ? (http://jimablett.hostwebs.com/)

Does it works ?

Hi Marco, I tried using Salvo's official 2.0.1 book, as well as several others (performance.bin, varied.bin), without success. The 64 bit compile works fine with the book, it's only the 32 bit compile that doesn't use the book.

[muxecoid]

I have funny degradation in some specific positions:

[d]r1bk1bnr/pp1p1ppp/3P1qn1/8/5P2/2N5/PPP1Q1PP/R1B1KB1R w KQ - 3 12

Here older Stockfish doesn't need much time to see that Ne4 is the best move with score around -1.0. Critter can spot it fast too. Stockfish 2.1 takes very long to find it and and even longer to give it score near -1.0

The tactical reason Ne4 is so good is that in many variations Ne4-g5-f7 forking maneuver is possible.

[bob]

Sorry to post any result below 20k games BTW FIDE Candidates Matches 2011 are 4 games each.

Jouni

They accept the error...

With computers, which don't get tired, doing so can be avoided...

[S.Taylor]

Wow really impressive so far. Stockfish 2.1 - Rybka 4.1 24,5 - 15,5
and Stockfish 2.1 - Critter 1.01 22,5 - 17,5 (2.0 scored 16 - 24 and 20 - 20 in same matches)! Surprisingly engine size is only 1/3 of 2.0 size.

Jouni

It is also a very good example of how little one should rely on such short matches for statistical evidence. Since there is no question R4 is stronger than Critter, the fact that it scored worse than Critter in parallel matches is helpful reinforcement.

The difference vs SF is quite small, and besides, it is more likely that SF was tested much more vs Rybka or Houdini, than vs other engines. So it may be optimized more vs Rybka than vs Critter.

[Albert Silver]

Sorry to post any result below 20k games BTW FIDE Candidates Matches 2011 are 4 games each.

Jouni

There is actually a big difference between the two. No one presumes that the winner of the match has shown himself to be statistically stronger. It just shows he won the match. It is a sports result.

[UncombedCoconut]

Thanks!

54 files changed, 4138 insertions(+), 5656 deletions(-)

So more then 1500 lines of code removed for this release.

Amazing... Is this mostly code simplification, or is it a lot of outright removal of features which don't increase ELO?

So far I see:
[ul][*]Re-merging of most remaining duplicated code (search() & root_search(), and code common to pawn & material-imbalance hashes)
[*]Removal of UCI knobs controlling search extensions
[*]Briefer opening book implementation
[*]Removal of "banner" comments, tighter text-wrap of comments, micro-reorganizations
[*]Header file reorganization[/ul]
[quote="muxecoid"The reported NPS is much lower than 2.0.1. Is it counted differently now?

Also eval numbers are wildly different now. Did stockfish become more materialistic?[/quote]
Doesn't look like nodes are counted differently. Which build are you using, and does it matter if you are using many threads?
This SF is slightly more "materialistic" in its search: it's more willing to reduce bad captures and futility-prune moves with negative SEE. Its eval looks nearly identical.

I can't seem to get the 32 bit Windows compile to work with a Polyglot book. Am I alone in my struggles?

Have you tried with the official Salvo's book ? (http://jimablett.hostwebs.com/)

Does it works ?

Hi Marco, I tried using Salvo's official 2.0.1 book, as well as several others (performance.bin, varied.bin), without success. The 64 bit compile works fine with the book, it's only the 32 bit compile that doesn't use the book.

Also works fine with the 32-bit Linux compile. Strange.