SmarThink
   

The results of some engines testing
SmarThink versions (download)
Opening books for the SmarThink (download)
Tournaments in which SmarThink has participated
SmarThink versions info
Thanks
Links (and shell for the SmarThink)
SmarThink details (NEW!)

The results of some engines testing

SmarThink versions
(New versions placed in the top of list)

Openings books for SmarThink (SMT)

Tournaments in which SmarThink has participated (PGN)

    SmarThink v0.17a[lpha] (01.10.2003)

    Changes:
    - fixed bug in king safety evalustion;
    - added evaluation for half-supported pawns;
    - added special evaluation for consequent promotions perspective;
    - improved interior-node recognition;
    - fixed minor bug in pawn races evaluation;
    - pawn advances evalustion changed;
    - additional knowledge for "pawns vs bishop" endgames;
    - changed evaluation for "good bishop";
    - check evasions generator improved;
    - move-ordering for check evasions and pawn promotions improved;
    - quiescence alhorythm improved;
    - added new quasy-Bitvinnik search extension;
    - added "one reply" extension;
    - added evalustion for out-quadrant pawns for non-pure pawn endgames;
    - changed evalustion of pawns on the bishop colour;
    - history-based pruning improved;
    - improved evaluation for "rook vs pawns" endgame;
    - several optimizations.
    The supposed ELO of this version (with DeepFritz openings book) is 2840 (+-30)

    SmarThink v0.16b[eta]++ (05.04.2003)

    Changes:
    - fixed minor bug in trapped rook evaluation;
    - fixed bug in piece centralization evaluation;
    - added new endgame knowledge.
    The supposed ELO of this version (with DeepFritz openings book) is 2794 (+-30)

    SmarThink v0.16b[eta]+ (18.03.2003)

    Changes:
    - evaluation and quiescence optimized (+5-7% speed up);
    - minor changes in endgame knowledge;
    - minor changes in passed pawns evaluation;
    - trapped rook evaluation added.
    The supposed ELO of this version (with DeepFritz openings book) is 2784 (+-30)

    SmarThink v0.16b[eta] (04.03.2003)

    Changes:
    - "perft" command implemented;
    - 2x32 bit hashkey replaced to 1x64 bit;
    - more aggressive EGTB probing;
    - fixed minor bug in quescence search;
    - some changes in endgame knowledge;
    - fixed bug in en passant capture hashing;
    - moves generator and make/uunmake optimized (3-5% speed up).
    The supposed ELO of this version (with DeepFritz openings book) is 2779 (+-30)

    SmarThink v0.16a[lpha]+ (13.02.2003)

    Changes:
    - fixed some bugs that was added in 0.16a;
    - new method of EGTB probe frame determining.
    The supposed ELO of this version (with DeepFritz openings book) is 2768 (+-30)

    SmarThink v0.16a[lpha] (11.02.2003)

    Changes:
    - added endgame tablebases (EGTB) support;
    - fixed minor bugs in endgame knowledge;
    - fixed minor bug in pawn structure evaluation;
    - added several endgame knowledge;
    - fixed trapped bishop evaluation;
    The supposed ELO of this version (with DeepFritz openings book) is 2766 (+-30)

    SmarThink v0.15b[eta]+ (31.01.2003)

    Changes:
    - fixed bugs in endgame knowledge;
    - changed evaluation of blocked pawns;
    - fixed bugs in analyze mode;
    The supposed ELO of this version (with DeepFritz openings book) is 2765 (+-30)

    SmarThink v0.15b[eta] (26.01.2003)

    Changes:
    - fixed minor bugs in endgame knowledge;
    - a few modifications in pawn structure evaluation;
    - additional rules for rook endings;
    - analyze mode (beta-version);
    The supposed ELO of this version (with DeepFritz openings book) is 2757 (+-30)

    SmarThink v0.15a[lpha] (22.01.2003)

    Changes:
    - fixed bugs with evaluation symmetry;
    - fixed bug in knight trajectory evaluation;
    - added several non well-known search extensions;
    - new game-staging algorythm;
    - new pruning ideology - "near-fine pruning";
    - new pruning technique for deep endgame;
    - moves generator and evaluation function optimized (search speed +15%);
    - added additional endgame knowledge.
    The supposed ELO of this version (with DeepFritz openings book) is 2755 (+-30)

    SmarThink v0.14a[lpha] (15.12.2002)

    Changes:
    - fixed bugs in handling some WB commands;
    - fixed bugs in evaluation function;
    - PV printing limit relaxed;
    - time management changed a little bit;
    The supposed ELO of this version (with DeepFritz openings book) is 2740 (+-30)

    SmarThink v0.13b[eta] (07.12.2002)

    Changes:
    - fixed minor bug in move generation;
    - chaged open files evaluation for rook endings;
    - changed several parts of pawn and knight endings evaluation;
    The supposed ELO of this version (with DeepFritz openings book) is 2725 (+-30)

    SmarThink v0.13a[lpha] (02.12.2002)

    Changes:
    - fixed bugs in some WinBoard commands processing;
    - fixed bugs in pondering;
    - added knight trajectories calculation for endgames;
    - weak pawns evaluation changed;
    - several optimizations.
    The supposed ELO of this version (with DeepFritz openings book) is 2720 (+-30)

    SmarThink v0.12a[lpha]+ (16.10.2002)

    Changes:
    - pondering support added;
    - bad trades code updated;
    - pawn endings evaluation updated (pawn races evaluation e.t.c.);
    - draw/half-draw positions recognizer for endgame fixed;
    - fixed bug with handling two or more commands in row during the search;
    - several optimizations.
    The supposed ELO of this version (with DeepFritz openings book) is 2715 (+-30)

    SmarThink v0.11b[eta]++ (20.09.2002)

    Changes:
    - little improvment in evaluation endgame with R vs R+P;
    - fixed immortal evil bug in time management.
    The supposed ELO of this version (with DeepFritz openings book) is 2697 (+-30)

    SmarThink v0.11b[eta]+ (19.09.2002)

    Changes:
    - fixed evil bug in time management;
    The supposed ELO of this version (with DeepFritz openings book) is 2695 (+-30)

    SmarThink v0.11b[eta] (19.09.2002)

    Changes:
    - fixed bug in PASSED_PAWN_PUSH extension;
    - SEE code optimized;
    The supposed ELO of this version (with DeepFritz openings book) is 2695 (+-30)

    SmarThink v0.11a[lpha]+ (17.09.2002)

    Changes:
    - fixed bug in time-checking.
    The supposed ELO of this version (with DeepFritz openings book) is 2650 (+-30)

    SmarThink v0.11a[lpha] (16.09.2002)

    Changes:
    - fixed several bugs in endgame evaluation (null-move in pawn endings, bishop+wrong pawn vs king);
    - fixed pawn hash and TT bugs;
    - futility pruning, ext. futility and razoring code changed;
    - history-based pruning code corrected;
    - attack-extensions on pre-leaf nodes added;
    - new search-extensions mechanism called "controlled singularity" added;
    - pawn structure and passed pawns evaluation is now more complicated;
    - material evaluation changed;
    - near-drawn endings (two different-color bishops e.t.c.) evaluation updated;
    - time control for multi-stage games fixed;
    - game-staging changed;
    - book learning added (see SmarThink.ini);
    - several optimizations.
    The supposed ELO of this version (with DeepFritz openings book) is 2650 (+-30)

    SmarThink v0.10a[lpha] (11.07.2002)

    Changes:
    - attack evaluation changed;
    - passed pawns evaluation changed;
    - added several positional knowledge;
    - several changes in search extension scheme;
    - several positional values optimized;
    - added "Draw by insufficient material" message.
    The supposed ELO of this version (with DeepFritz openings book) is 2620 (+-30)

    SmarThink v0.09a[lpha] (02.06.2002)

    Changes:
    - learning added (see SmarThink.ini);
    - passed pawn chains evaluation added;
    - developing evaluation changed;
    - attack evaluation changed;
    - bishop/knight evaluation changed;
    - mobility evaluation changed.
    The supposed ELO of this version (with DeepFritz openings book) is 2600 (+-30)

    SmarThink v0.08b[eta] (11.05.2002)

    Changes:
    - fixed bug in king pawn shield evaluation;
    - fixed bug in resigning;
    - added rook to 5 pawns trade evaluation;
    The supposed ELO of this version (with DeepFritz openings book) is 2585 (+-30)

    SmarThink v0.08a[lpha] (11.05.2002)

    Changes:
    - added resign feature (see ResignValue and ResignTimes values in SmarThink.ini);
    - fixed bug in attacking/defending middlegame evaluation;
    - added knight outpost evaluation;
    - added several endgame knowledge;
    - added metarial-dependent king safety and passed pawns evaluation;
    - several small optimizations;
    The supposed ELO of this version (with DeepFritz openings book) is 2555 (+-30)

    SmarThink v0.07a[lpha] (05.05.2002)

    Changes:
    - general bugfixing;
    - added some strategical knowledge;
    - several small optimizations;
    The supposed ELO of this version (with DeepFritz openings book) is 2560 (+-30)

    SmarThink v0.06b[eta] (03.05.2002)

    Changes:
    - added several endgame knowledge;
    - fixed bug in pawn endings evaluation;
    - added bad trades evaluation;
    - several small optimizations (3-8% search speed growth);
    The supposed ELO of this version (with DeepFritz openings book) is 2540 (+-30)

    SmarThink v0.06a[lpha] (27.04.2002)

    Changes:
    - added special endgame knowledge;
    - attack evaluation for middlegame modified;
    - parameters "Aggressivity" and "Cowardice" now added in SmarThink.ini;
    - several small optimizations;
    The supposed ELO of this version (with DeepFritz openings book) is 2530 (+-30)

    SmarThink v0.05a[lpha] (10.04.2002)

    Changes:
    - several changes in null-move pruning;
    - added attack evaluation for middlegame;
    - time management changed a little;
    - several small optimizations;
    The supposed ELO of this version (with DeepFritz openings book) is 2515 (+-30)

    SmarThink v0.04a[lpha] (27.03.2002)

    Changes:
    - several changes in null-move pruning;
    - added game-stage dependent passed pawns evaluation;
    - added blocked passed pawns evaluation;
    - added pins evaluation;
    - changed evaluation for bishop and bishop pair;
    - time management changed;
    - several small optimizations;
    The supposed ELO of this version (with DeepFritz openings book) is 2500 (+-30)

    SmarThink v0.03b[eta] (07.03.2002)

    Changes:
    - several changes in null-move pruning;
    - SmarThink now compiled using Intel C/C++ 5.0;
    - several small optimizations;
    The supposed ELO of this version (with DeepFritz openings book) is 2455 (+-30)

    SmarThink v0.03a[lpha] (10.02.2002)

    Changes:
    - changes in pruning scheme;
    - "smart" null-move pruning added;
    The supposed ELO of this version (with DeepFritz openings book) is 2445 (+-30)

    SmarThink v0.02b[eta] (27.01.2002)

    Changes:
    - changes in pruning scheme;
    - fixed bug in threefold repetition detection;
    The supposed ELO of this version (with DeepFritz openings book) is 2435 (+-30)

    SmarThink v0.01a[lpha]+ (25.01.2002)

    Changes:
    - fixed evaluation corrupting when search stops;
    The supposed ELO of this version (with DeepFritz openings book) is 2425 (+-30)

    SmarThink v0.01a[lpha] (25.01.2002)

    Changes:
    - SmarThink now retrieves PV from transposition table (10% faster);
    - fixed bugs in hashing scheme;
    - several small optimizations.

    The supposed ELO of this version (with DeepFritz openings book) is 2410 (+-30)

    SmarThink v0.01b[eta]+ (29.12.2001)

    Changes:
    - king pawn shield evaluation changed which causes some search speed growth and better scoring;
    - fixed bug in castling state calculation;
    - several small optimizations.

    The supposed ELO of this version (with DeepFritz openings book) is 2415 (+-30)

    SmarThink v0.01b[eta] (26.12.2001)

    Changes:
    - fixed a bug in open files calculation;
    - futility pruning completed;
    - new method of tree reducing (ordering based pruning) realized;
    - used "smart" method of mobility calculation;

    - several small optimizations.

    The supposed ELO of this version (with DeepFritz openings book) is 2412 (+-30)

    SmarThink v0.01a[lpha]+ (18.10.2001)

    Changes:
    - fixed a little bug in heuristic calculation;
    - added "selective cuttings" to drop some redundant nodes in search tree;
    - several small optimizations

    The supposed ELO of this version (with DeepFritz openings book) is 2395 (+-30)



    SmarThink v0.01a[lpha] (18.10.2001)

    Changes:
    - Openings book added;
    - "bkm" command added; this command prints the list of moves from openings book for the current position;
    - The names of versions now have a short form; E.G. v0.01alpha = v0.01a

    The supposed ELO of this version (with DeepFritz openings book) is 2385 (+-30)

Links

Recommended shell for the SmarThink (Arena)
Russian computer chess site (russian)
WinBoard forum (english)
Site for the chess programming beginners (russian)