Marek Kwiatkowski wrote:Sting SF is open source code software (thread) based on Stockfish (2.1.1) without exe-files.
Sting SF is to help detect, impossible to find or time consuming for engines, sacrifices.
Unfortunately, but just Stockfish is very weak to find difficult sacrifices and thus is not a good tool to make any analysis.
Sting SF project should very help you to get a reliable result in analysis.
Sting SF uses the option : “Sacrifice detection” (0 – 8, 8 = default, 0 = disabled).
BTW. I tested a strength (Sacrifice detection = 8). Sting seems to be stronger than Stockfish.

www.free.of.pl/c/ccintorun/g/sting.rar

The difference in the code - my idea - is below:

Code:

// Step extra (after step 7) - detection of good sacrifices.
if (    depth >= 8 * ONE_PLY       
        &&  SacrificeDetection // default option = 8 
        && !ss->skipSacrificeDetection
        &&  pos.captured_piece_type() != PIECE_TYPE_NONE
        && !Root
        &&  beta < VALUE_KNOWN_WIN)
    {
      Value q;
      if ((ss-1)->captureValue != VALUE_NONE) q = -(ss-1)->captureValue;
      else if (tte && tte->depth() >= 8 * ONE_PLY) q = ss->eval;
      else q = ss->eval = qsearch<PV>(pos, ss, -VALUE_INFINITE, VALUE_INFINITE, DEPTH_ZERO));
     
      TT.store(posKey, VALUE_NONE, VALUE_TYPE_NONE, DEPTH_NONE, ttMove, ss->eval, ss->evalMargin);
             
      Depth d = depth - depth/10;
      int i = 0;
      Value step = PawnValueMidgame / SacrificeDetection;
      Value v = q;

      while (true)
      {
        ss->captureValue = v;     
        Value rAlpha = Min(alpha, v + step);
        if (rAlpha == alpha && d <= depth) break;
        if (d >= depth + depth/10) rAlpha = alpha;
        Value rBeta = rAlpha == alpha ? beta : rAlpha + 1;

        ss->skipSacrificeDetection = true;       
        v = search<PvNode>(pos, ss, rAlpha, rBeta, d);
        ss->skipSacrificeDetection = false;       
     
        i++;

        if (v > rAlpha && d >= depth)
        {
          if (v > alpha) return v;
          if (v >= PawnValueMidgame + step)
          {
            Value bonus = Min(QueenValueMidgame, (v - q) / step * (i * i * SacrificeDetection));           
            if (v + bonus > alpha)
            {
              TT.store(posKey, VALUE_NONE, VALUE_TYPE_NONE, DEPTH_NONE, ss->bestMove, ss->eval, ss->evalMargin);
              return v + bonus;
            }         
          }       
        }
        if (v <= rAlpha) return v;       

        d = d + ONE_PLY;       
      }     
    }

Code:

Games Completed = 500 of 500 (Avg game length = 30.507 sec)
Settings = RR/32MB/4000ms+200ms/M 700000cp for 7 moves, D 150 moves/EPD:E:\LittleBlitzer\openings.epd(4000)
Time = 3916 sec elapsed, 0 sec remaining

 1.  Stockfish 2.1.1 JA         290.5/500   168-87-245     (L: m=87 t=0 i=0 a=0)   (D: r=171 i=42 f=19 s=2 a=11)   (tpm=232.8 d=14.2 nps=1018406)
 2.  Sting SF 111208             209.5/500   87-168-245     (L: m=168 t=0 i=0 a=0)   (D: r=171 i=42 f=19 s=2 a=11)   (tpm=235.7 d=13.4 nps=945812)


Code:

Games Completed = 500 of 500 (Avg game length = 30.694 sec)
Settings = RR/32MB/4000ms+200ms/M 700000cp for 7 moves, D 150 moves/EPD:E:\LittleBlitzer\openings.epd(4000)
Time = 3950 sec elapsed, 0 sec remaining

 1.  Stockfish 2.1.1 JA         295.5/500   171-80-249     (L: m=80 t=0 i=0 a=0)   (D: r=189 i=34 f=19 s=1 a=6)   (tpm=231.9 d=14.2 nps=1016178)
 2.  Sting SF JA 1.0             204.5/500   80-171-249     (L: m=171 t=0 i=0 a=0)   (D: r=189 i=34 f=19 s=1 a=6)   (tpm=237.3 d=13.4 nps=967757)