new epd test suite

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Dann Corbit, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Hart

new epd test suite

Post by Hart » Sun Oct 18, 2009 1:20 am

I designed this suite to map a "rated time" as determined through "automatic analysis" in Arena, to a CEGT 40/20 rating and from my results so far it mostly works. I ran six tests with it when I was finished and it maps these engines (x64) "rated times" to these ratings, after a total of 15 minutes of analysis:

Naum 4 t1: 3068
Naum 4 t2: 3083
Naum 4 t4: 3105

Rybka 3 t1: 3116
Rybka 3 t2: 3145
Rybka 3 t4: 3197

Naum's scores map very well to it respective CEGT ratings and Rybka 3 t1 scores just a hair over Naum 4 t4 as well which is what we would expect if their CEGT ratings are correct.

To design the suite, I ran five engines over 658 positions I ripped from high level ICCF games, each for 10 seconds and I then discarded all positions where all engines failed the position with a time of 10 seconds or all solved it easily in under 1 second.

I then shuffled the positions randomly 2x10^7 times until I could take the total times for each engine, for the first 30 positions, and they matched up very well their CEGT ratings through a logarithmic regression. The code that tells me when my positions are usable:

Code: Select all

double LogError(double x[]) {

	int 	i, j;
	double 	
			diffTime,
			diffRating,
			sum = 0.0;
	
	for &#40;i=0; i<ENGINES-1; i++) 
		for &#40;j=i+1; j<ENGINES; j++) &#123;

			diffRating = ratings&#91;i&#93; - ratings&#91;j&#93;;
			diffTime = log&#40;x&#91;j&#93; / x&#91;i&#93;)/log&#40;2.0&#41;;
			
			sum += pow&#40;diffRating - diffTime*SPEEDFACTOR,2&#41;;
		&#125;

	return sqrt&#40;sum/(&#40;ENGINES-1&#41;*&#40;ENGINES/2&#41;));
&#125;
Where ratings[0,...,ENGINES-1], are the respective ratings (high to low) of the five highest rated engines at CEGT 40/20, and ENGINES equals five. Worth noting is that I got optimal results when SPEEDFACTOR equaled 55, and for weaker engines on the CEGT list like Fruit, Yace, SOS, a SPEEDFACTOR of 60 gave the best results. Where SPEEDFACTOR is equal to the Elo increase an engine should gain for solving positions in half time. The function computes the rating differences between all ratings for every engines and then it calculates what these differences should be given the differences in the times it took for every engine to solve the 30 positions. The smaller the error returned, the greater the logarithmic regression between engines ratings and their times.

For example, Rybka solves these positions with a rated time of ~34, and Shredder ~122. This theoretically equates to a Elo difference of 55*log(122/34)/log(2)=101. The difference between their CEGT ratings is 112. The error here is 112 - 101 = 11.

The suite is tuned for engines rated between 3200 and 3000 (CEGT) at 10 seconds of analysis on a Q6600 3.0GHz. YMMV for different engines on different hardware. It becomes difficult to test engines rated more than 200 Elo apart due to the lack of precision in Arena's "rated times", which are whole numbers 0 through 10, in this case. For engines 330 Elo apart we'd expect a 2^(330/55) = 64x difference in time which would require we test every 658 position with every engine for a minimum of 64 seconds. This can only be done with days of analysis time, which i can't afford. But that would be the ideal way to go. For Rybka, Naum, Shredder, Stockfish, and Zappa II (all x64 and t=4), this function maps their "rated time" to their CEGT ratings with R^2 = .9984 :

RATING = 3513 + -89.8 ln x
where x is the rated time.

The methodology still needs some refinement but I think I am on the right track.

Code: Select all

2r5/pp1bkpp1/2rnp2p/3p4/3P4/P2BP1P1/1P1KNPP1/2R4R w - - fmvn 22; hmvc 13; bm Rxc6;
r2b2k1/pp6/4rp1p/2p5/4P3/5N2/PP4PP/3R1RK1 w - - fmvn 22; hmvc 0; bm Nh4;
7r/1p1kbppp/p1b1p3/4P3/5P2/P3B3/1P2BKPP/2R5 w - - fmvn 22; hmvc 2; bm Bc5;
r5k1/pp4pp/2p5/3rNb2/3P3P/P7/1P4P1/3RR1K1 w - - fmvn 22; hmvc 3; bm b4;
r2r2k1/pq3pb1/2p1n1p1/2B1p1Pp/4N2P/5P2/PPP2Q2/2KRR3 w - - fmvn 22; hmvc 0; bm Bd6;
2rq2k1/1p4pp/3p2r1/3Ppp2/bP6/P5P1/2PQBP1P/R3K2R w KQ - fmvn 22; hmvc 0; bm c4;
4qr1k/3b2pp/p1rpp3/1p5n/4P1B1/2N1QP2/PPP4P/1K1R2R1 w - - fmvn 22; hmvc 3; bm e5;
2kr3r/1b3p2/pq2p2p/2b1P1p1/PpB3Q1/1Pn5/5PPP/R1B1NRK1 w - - fmvn 22; hmvc 0; bm Qh5;
r7/1b1nkp1p/1r1ppp1b/8/1P2PP2/3Q2P1/P1P4P/2KRR3 w - - fmvn 22; hmvc 1; bm a3;
3r2k1/1ppr1qpp/2n1b3/p2np3/8/PP1P1NP1/1BQ1PKBP/2RR4 w - - fmvn 22; hmvc 0; bm Qc5;
3rr1k1/pb3ppp/8/1p1n4/8/PB3N2/1P3PPP/3R1RK1 w - - fmvn 22; hmvc 2; bm Rfe1;
2r2rk1/3pb1pp/p2npP2/7P/1p1NPP2/3qB3/PPP5/1K1R1R2 w - - fmvn 22; hmvc 0; bm cxd3;
4k3/1r2b2p/p1q1pnr1/2p5/1pp2B2/2N2PPQ/PP2P2P/R2R2K1 w - - fmvn 22; hmvc 0; bm Ne4;
6k1/pq2ppb1/1p4pp/n7/3P3B/3QPN1P/P4PP1/6K1 w - - fmvn 22; hmvc 0; bm Qb5;
r4r1k/1p3pp1/3b1n2/p2p4/3N4/P1B1P2q/1P2QP1P/R3K1R1 w Q - fmvn 22; hmvc 1; bm Qf3;
r1br2k1/1p1n1pp1/p1nBp2p/q7/4N3/6Q1/2PRB1PP/4K2R w K - fmvn 22; hmvc 7; bm Bc7;
3rr1k1/ppp2ppp/8/3P1b2/2Pq4/PQ6/4BPPP/R3R1K1 w - - fmvn 22; hmvc 0; bm Bf1;
rn3rk1/pppb4/3p2pp/3P4/1PP1R3/P4BP1/1b3P1P/4Q1K1 w - - fmvn 22; hmvc 0; bm Qe3;
5rk1/p5p1/1ppr3p/1Q1nqp2/1PB5/P2RP3/5PPP/5RK1 w - - fmvn 22; hmvc 0; bm Qa4;
2b1rr1k/5ppp/2p2n2/ppq1PPB1/2p3P1/2P5/P3Q1BP/4RR1K w - - fmvn 22; hmvc 0; bm Bh4;
3r1r2/1p3pkp/p1bRp3/4Pp1B/8/1NN5/P1P3PP/6K1 w - - fmvn 22; hmvc 1; bm Rd1;
1q1r2k1/p2nbppp/b1p1p3/8/P1pPP3/2B3P1/2QN1PBP/3R2K1 w - - fmvn 22; hmvc 2; bm Ba1;
r3r3/pp3pk1/2b2qp1/7p/5n1P/2N3Q1/PP3PP1/R2R1BK1 w - - fmvn 22; hmvc 1; bm f3;
1r3r1k/3q2pp/B2p4/1P1Nppb1/2Q1P3/8/1P3PPP/5RK1 w - - fmvn 22; hmvc 0; bm exf5;
r3kb1r/2qn1ppp/p3pn2/5N2/2P5/2P1BP2/1N2Q1PP/R2R2K1 w kq - fmvn 22; hmvc 7; bm Na4;
r4r2/pp3pk1/2b1p1p1/P2pP2p/3P3P/3B3R/2PK1PP1/R7 w - - fmvn 22; hmvc 0; bm Rb1;
4r1k1/5pp1/3b4/1p1p1b2/3P3p/2P1N1Pq/1P3P1P/R1BQ2K1 w - - fmvn 22; hmvc 0; bm Qf1;
r1r3k1/4qpbp/2bp1np1/ppn1p3/2P1P3/1P2BNNP/P1BQ1PP1/2R1R1K1 w - - fmvn 22; hmvc 0; bm Bxc5;
3rr3/p1qn1kp1/1p3b1p/8/3pQ3/1P3N2/PB3PPP/3R1RK1 w - - fmvn 22; hmvc 0; bm Qd5+;
r6r/ppp2kp1/1b1qb2p/5p2/8/5NN1/PPQ2PPP/R4RK1 w - - fmvn 22; hmvc 6; bm Rfe1;

Hart

Re: new epd test suite

Post by Hart » Mon Oct 19, 2009 1:03 pm

Here is the revised suite with 30 more positions. I reran all tests with 1-CPU engines and my results are far more consistent now. The inconsistency in the MP engines was making it much more difficult to get reliable results. If you do use these positions to test engines using MP's, you'll probably need to take the average of several trials to get reliable results.
My coefficient of determination is still high at .993, with Rating = 3558 + -89 ln x.
Suggested usage: to gauge the difference between the strength of two programs, run these positions with enough time so that you're solving them in approximately half of the total rated time. For engine A to solve in half the time as engine B should equate to a difference in Elo of 55 on faster hardware with stronger engines or 65 Elo for weaker engines on slower hardware.

Code: Select all

3r2k1/2pr3p/1pR2pn1/p7/3p4/Pb1P1NP1/4P1BP/4R1K1 b - - fmvn 30; hmvc 1; bm Bd5;
5rk1/1p4bp/3Pq3/2P1p3/5P2/3P1PP1/1Q4KP/5R2 b - - fmvn 30; hmvc 0; bm Qd5;
rr4k1/2q1bppp/3p4/p2Rn1P1/1p1NPQ1P/8/PPP5/1K3B1R w - - fmvn 22; hmvc 2; bm h5;
r2qnrk1/4bppp/3p4/4nPP1/4P3/Np2B3/1P1Q3P/1K1R1B1R w - - fmvn 22; hmvc 2; bm h4;
5rk1/7p/p2N2p1/2pP1b2/p7/7P/1P3PP1/4R1K1 b - - fmvn 30; hmvc 0; bm Rb8;
3q3k/1br3r1/1ppn1pp1/3p3p/3P1P2/1P2P3/2Q1NPBP/R1R3K1 b - - fmvn 30; hmvc 1; bm h4;
1rbr1k2/p1n1qpbB/7p/1ppPp1p1/4Q3/P1N2NP1/1P3PP1/3RR1K1 w - - fmvn 22; hmvc 0; bm Nxe5;
8/p1p2rk1/1p1pq1pb/1P1Qp2p/P1P4P/5P2/5PK1/4R2R b - - fmvn 30; hmvc 4; bm Qf5;
b1r3k1/3q1ppp/1p2pn2/pPb1n3/3NP3/PN4P1/1BQ2PBP/2R3K1 w - - fmvn 22; hmvc 7; bm a4;
1r4k1/3q2pp/pp2p3/n1p1p1B1/2P1P3/P1P4P/4QPPK/1R6 b - - fmvn 30; hmvc 4; bm h6;
4r1k1/5p1p/2r2p2/p7/1p3P2/4p1PP/P3P1BK/1R4N1 b - - fmvn 30; hmvc 1; bm Rc5;
2br3k/3q1pbB/pr1P4/1p2p3/8/N2Q4/PP3PPP/3R1RK1 w - - fmvn 22; hmvc 1; bm Qe4;
1n4k1/2q1bppp/p1p1p3/2Pp4/3P4/2BQP2P/P3NPP1/6K1 w - - fmvn 22; hmvc 0; bm Nc1;
3qr1k1/pp6/6b1/3PpnB1/2r5/2N4P/PP1Q2B1/3R3K b - - fmvn 30; hmvc 3; bm Qb6;
2r1r2k/ppBb1pp1/2n4p/3q4/3p4/P2B2Q1/1PP2PPP/R3R1K1 w - - fmvn 22; hmvc 12; bm Qd6;
r2k1b2/p2n1p2/7p/2pR4/Nppr3P/6B1/PP3PP1/2KR4 w - - fmvn 22; hmvc 5; bm R1xd4;
8/pp3kp1/2n1b2p/5p2/5P2/P7/1PBB1KPP/8 b - - fmvn 30; hmvc 0; bm Nd4;
r5k1/1pb1q1pp/p3p3/P2p2n1/3B2n1/1N1Q2P1/1PP2P1P/R4NK1 w - - fmvn 22; hmvc 0; bm f4;
8/1p3pk1/p2p2q1/3n4/2PNbP2/8/P1PQ2p1/2K3R1 b - - fmvn 30; hmvc 1; bm Nf6;
2r1Rk2/p4pp1/7p/3p2n1/3N4/1P3P2/P1P3P1/5RK1 b - - fmvn 30; hmvc 0; bm Kxe8;
r3r1k1/pp1b1pp1/5n1p/1NqP4/1b2P3/1PPQ3P/P2N1PP1/4RRK1 w - - fmvn 22; hmvc 1; bm Nc7;
r5k1/3p1pp1/4rn1p/q1p1p3/N1PnP2B/P4P2/1P1R2PP/3QK2R w K - fmvn 22; hmvc 1; bm Nc3;
2q2rk1/p4p1p/1b2p1p1/5n2/1p2NP2/4P3/PP1Q1PBP/5RK1 w - - fmvn 22; hmvc 1; bm Qxb4;
8/p1r2pk1/1p1R1qp1/4p2p/2P4P/1P4P1/P4PK1/3Q4 b - - fmvn 30; hmvc 3; bm Qf5;
2b1rr1k/5ppp/2p2n2/ppq1PPB1/2p3P1/2P5/P3Q1BP/4RR1K w - - fmvn 22; hmvc 0; bm Bh4;
1r3rk1/3q1pp1/R1npb2p/3NpNbP/4P3/2pB4/1P3PP1/3Q1RK1 w - - fmvn 22; hmvc 0; bm bxc3;
1r2r1k1/p2b1p2/2p3p1/q2n1pNp/NbBP3P/1P2P1P1/1P1Q1P2/1KRR4 w - - fmvn 22; hmvc 9; bm Qe2;
6k1/1b3p1p/pq1r1np1/1p6/1N3Q2/5P1P/PPB3P1/3R3K b - - fmvn 30; hmvc 1; bm Rxd1+;
2r2rk1/3R1pp1/7p/4p2P/4Q3/1Pq2P2/P7/1K1R4 b - - fmvn 30; hmvc 1; bm Rfe8;
2r1r1k1/pp2qppp/1n2b3/8/3PpNP1/1P2PB1P/P2Q4/2R1R1K1 w - - fmvn 22; hmvc 0; bm Rxc8;
r1n2rk1/1p3ppp/p1n1b3/2P1p3/4P3/2P1BP2/1N1R2PP/R4BK1 b - - fmvn 30; hmvc 18; bm Rd8;
1r2r1k1/1q1bbppp/5n2/1p2p3/p3P3/3BB2P/PP1NQPP1/2R1R1K1 w - - fmvn 22; hmvc 0; bm Nf3;
8/3k4/2p3p1/p1Bnr3/Pp2p2P/1P4P1/2PK4/4R3 b - - fmvn 30; hmvc 0; bm Re8;
6k1/pq2ppb1/1p4pp/n7/3P3B/3QPN1P/P4PP1/6K1 w - - fmvn 22; hmvc 0; bm Qb5;
7r/3k2bp/p7/8/3p1B2/1Pp5/P1P2KPP/3R4 b - - fmvn 30; hmvc 5; bm Rf8;
3rr1k1/1p2bppp/pn6/8/8/2P2N2/PP3PPP/R1B1R1K1 w - - fmvn 22; hmvc 4; bm b3;
3r3k/6b1/R1n2r1p/1q1p2pB/4p2P/2P3B1/5PP1/R1Q3K1 b - - fmvn 30; hmvc 0; bm d4;
2r5/p4pk1/3Qbqp1/1p2p2p/2r1P2P/1NP2P2/PP4P1/1K1R3R w - - fmvn 22; hmvc 0; bm Rd2;
1r4k1/1bq1bppp/3p1n2/p1n1pP2/4P1P1/1P1B3Q/NP1B3P/1K1RR3 w - - fmvn 22; hmvc 1; bm Nc3;
2r3k1/p1pqrpp1/1pn2b1p/3P4/2P2B2/P2Q1N1P/5PP1/1R2R1K1 w - - fmvn 22; hmvc 1; bm Rbd1;
3r1r2/1p3pkp/p1bRp3/4Pp1B/8/1NN5/P1P3PP/6K1 w - - fmvn 22; hmvc 1; bm Rd1;
3rr1k1/pb3ppp/8/1p1n4/8/PB3N2/1P3PPP/3R1RK1 w - - fmvn 22; hmvc 2; bm Rfe1;
1r3rk1/2q3pp/3b1p2/p1pPp3/PpQ5/5P2/1PP1RBPP/R5K1 w - - fmvn 22; hmvc 0; bm b3;
r3r3/pp3pk1/2b2qp1/7p/5n1P/2N3Q1/PP3PP1/R2R1BK1 w - - fmvn 22; hmvc 1; bm f3;
3r2k1/1ppr1qpp/2n1b3/p2np3/8/PP1P1NP1/1BQ1PKBP/2RR4 w - - fmvn 22; hmvc 0; bm Qc5;
bqr1r1k1/5pbp/pp1ppnp1/2n5/2P1P3/1PN2PP1/PBN1R1BP/3RQ2K w - - fmvn 22; hmvc 5; bm Qd2;
r1k2b1r/2q2p2/p2p1P2/2nP3p/1p1N2b1/8/PPP2Q1P/1K2RBR1 w - - fmvn 22; hmvc 5; bm h3;
1r4k1/1b2bppp/3p4/p4P2/N2RP3/4R3/1PB4n/1K6 b - - fmvn 30; hmvc 1; bm Bc6;
5qk1/1prn2pp/p1r1pb2/4n3/3QPB2/P1N3PP/1PP5/1K1R1B1R w - - fmvn 22; hmvc 1; bm Qb4;
2r2rk1/1n1qb1pp/pn6/2pPp3/p3N3/2P1BN1P/5PP1/R2QR1K1 w - - fmvn 22; hmvc 0; bm Rb1;
r3r1k1/3p1ppp/2b2n2/4n1B1/Pq2P3/2N3PP/3Q2B1/4RRK1 w - - fmvn 22; hmvc 1; bm Bxf6;
5rk1/1p1qb2p/p2p1pp1/P2Pn3/1PrN4/1Q6/6PP/1R3RBK b - - fmvn 30; hmvc 5; bm Qg4;
2rqrnk1/3bbpp1/p6p/3p4/1n1P1PP1/2N5/1P2N1QP/RBB3RK w - - fmvn 22; hmvc 1; bm g5;
2r2rk1/pq1n1pp1/4pn1p/1pp5/3P3B/P3PP2/1P2Q1PP/1N1R1RK1 w - - fmvn 22; hmvc 0; bm Nc3;
r4bk1/5pp1/p1b1qn1p/2p1B3/Pp1P4/4N3/1PB2PPP/R2Q2K1 w - - fmvn 22; hmvc 4; bm Rc1;
5r1k/pp2pq1p/2pr1np1/5P2/3P3P/P3P1QB/1P5K/2R3R1 b - - fmvn 30; hmvc 0; bm Qg7;
2q3k1/p1p2p1p/2p3p1/8/4R3/1PQ1P2P/P4PPK/3r4 b - - fmvn 30; hmvc 1; bm Rd6;
3r1r2/6kp/8/1R3p2/3b1p2/8/3N1PPP/5RK1 b - - fmvn 30; hmvc 2; bm Rb8;
3r2k1/5pp1/2n1p2p/2p1P3/n3NP2/PpBr4/1P2RKPP/2R5 b - - fmvn 30; hmvc 2; bm c4;
4qr1k/3b2pp/p1rpp3/1p5n/4P1B1/2N1QP2/PPP4P/1K1R2R1 w - - fmvn 22; hmvc 3; bm e5;

Dann Corbit
Posts: 12040
Joined: Wed Mar 08, 2006 7:57 pm
Location: Redmond, WA USA
Contact:

Re: new epd test suite

Post by Dann Corbit » Tue Oct 20, 2009 7:46 pm

Here is my current analysis for this suite. I gave Rybka only 10 minutes per position with 4 CPUs at 3GHz and 1.2 GB RAM. Some Rybka answers are different, and several positions also seem to be toss-up (more than one move has about the same value).

Code: Select all

3r2k1/2pr3p/1pR2pn1/p7/3p4/Pb1P1NP1/4P1BP/4R1K1 b - - bm Bd5; id "Michael Hart.001"; c0 "Bd5=10, Ba4=5, Ne7=4, Re7=8";
5rk1/1p4bp/3Pq3/2P1p3/5P2/3P1PP1/1Q4KP/5R2 b - - bm Rf7; id "Michael Hart.002"; c0 "Rf7=10, b6=5, exf4=1";
rr4k1/2q1bppp/3p4/p2Rn1P1/1p1NPQ1P/8/PPP5/1K3B1R w - - bm h5; id "Michael Hart.003"; c0 "h5=10, Bb5=10, Nb5=10, Qd2=10";
r2qnrk1/4bppp/3p4/4nPP1/4P3/Np2B3/1P1Q3P/1K1R1B1R w - - bm Rc1; id "Michael Hart.004"; c0 "Rc1=10, Bf4=10, Qg2=10, h4=10";
5rk1/7p/p2N2p1/2pP1b2/p7/7P/1P3PP1/4R1K1 b - - bm Rf6; id "Michael Hart.005"; c0 "Rf6=10, Rb8=4, Rd8=5, h5=4";
3q3k/1br3r1/1ppn1pp1/3p3p/3P1P2/1P2P3/2Q1NPBP/R1R3K1 b - - bm Kh7; id "Michael Hart.006"; c0 "Kh7=10, Nf5=5, Qc8=3, h4=3";
1rbr1k2/p1n1qpbB/7p/1ppPp1p1/4Q3/P1N2NP1/1P3PP1/3RR1K1 w - - bm b4; id "Michael Hart.007"; c0 "b4=10, Nxe5=3, Qc2=6, Rd2=6";
8/p1p2rk1/1p1pq1pb/1P1Qp2p/P1P4P/5P2/5PK1/4R2R b - - bm Qe7; id "Michael Hart.008"; c0 "Qe7=10, Qd7=3, Qe8=3, Qf5=3";
b1r3k1/3q1ppp/1p2pn2/pPb1n3/3NP3/PN4P1/1BQ2PBP/2R3K1 w - - bm Ba1; id "Michael Hart.009"; c0 "Ba1=10, Nc6=10, Qe2=10, a4=10";
1r4k1/3q2pp/pp2p3/n1p1p1B1/2P1P3/P1P4P/4QPPK/1R6 b - - bm Qc7; id "Michael Hart.010"; c0 "Qc7=10, Kh8=5, Qc6=4, h6=1";
4r1k1/5p1p/2r2p2/p7/1p3P2/4p1PP/P3P1BK/1R4N1 b - - bm Rc3; id "Michael Hart.011"; c0 "Rc3=10, Rc2=10, Rc5=10, Rd6=10";
2br3k/3q1pbB/pr1P4/1p2p3/8/N2Q4/PP3PPP/3R1RK1 w - - bm Qe3; id "Michael Hart.012"; c0 "Qe3=10, Qb1=10, Qc2=10, Qe4=10";
1n4k1/2q1bppp/p1p1p3/2Pp4/3P4/2BQP2P/P3NPP1/6K1 w - - bm Be1; id "Michael Hart.013"; c0 "Be1=10, Bd2=10, Nc1=10, f3=10";
3qr1k1/pp6/6b1/3PpnB1/2r5/2N4P/PP1Q2B1/3R3K b - - bm Qc8; id "Michael Hart.014"; c0 "Qc8=10, Qb6=10, Qd6=10, Qd7=10";
2r1r2k/ppBb1pp1/2n4p/3q4/3p4/P2B2Q1/1PP2PPP/R3R1K1 w - - bm Rad1; id "Michael Hart.015"; c0 "Rad1=10, Qd6=10, Qf4=10, Rxe8+=10";
r2k1b2/p2n1p2/7p/2pR4/Nppr3P/6B1/PP3PP1/2KR4 w - - bm Bd6; id "Michael Hart.016"; c0 "Bd6=10, Kb1=6, R1xd4=10, R5xd4=10";
8/pp3kp1/2n1b2p/5p2/5P2/P7/1PBB1KPP/8 b - - bm a6; id "Michael Hart.017"; c0 "a6=10, Nd4=5, Ne7=5, g6=5";
r5k1/1pb1q1pp/p3p3/P2p2n1/3B2n1/1N1Q2P1/1PP2P1P/R4NK1 w - - bm f4; id "Michael Hart.018"; c0 "f4=10, Qe2=3, Re1=2, f3=5";
8/1p3pk1/p2p2q1/3n4/2PNbP2/8/P1PQ2p1/2K3R1 b - - bm Nc7; id "Michael Hart.019"; c0 "Nc7=10, Nb6=10, Ne7=10, Nf6=10";
2r1Rk2/p4pp1/7p/3p2n1/3N4/1P3P2/P1P3P1/5RK1 b - - bm Kxe8; id "Michael Hart.020"; c0 "Kxe8=10, Rxe8=7";
r3r1k1/pp1b1pp1/5n1p/1NqP4/1b2P3/1PPQ3P/P2N1PP1/4RRK1 w - - bm Nd6; id "Michael Hart.021"; c0 "Nd6=10, Nc4=5, Nc7=2, cxb4=4";
r5k1/3p1pp1/4rn1p/q1p1p3/N1PnP2B/P4P2/1P1R2PP/3QK2R w K - bm Nxc5; id "Michael Hart.022"; c0 "Nxc5=10, Bxf6=1, Nc3=1, b3=5";
2q2rk1/p4p1p/1b2p1p1/5n2/1p2NP2/4P3/PP1Q1PBP/5RK1 w - - bm b3; id "Michael Hart.023"; c0 "b3=10, Nf6+=10, Qxb4=10, h3=10";
8/p1r2pk1/1p1R1qp1/4p2p/2P4P/1P4P1/P4PK1/3Q4 b - - bm Rc6; id "Michael Hart.024"; c0 "Rc6=10, Qe7=1, Qf5=1, Re7=5";
2b1rr1k/5ppp/2p2n2/ppq1PPB1/2p3P1/2P5/P3Q1BP/4RR1K w - - bm Bh4; id "Michael Hart.025"; c0 "Bh4=10, Bf4=8, a4=8, h3=8";
1r3rk1/3q1pp1/R1npb2p/3NpNbP/4P3/2pB4/1P3PP1/3Q1RK1 w - - bm Qf3; id "Michael Hart.026"; c0 "Qf3=10, Rxc6=5, bxc3=7";
1r2r1k1/p2b1p2/2p3p1/q2n1pNp/NbBP3P/1P2P1P1/1P1Q1P2/1KRR4 w - - bm Nc3; id "Michael Hart.027"; c0 "Nc3=10, Qc2=10, Qd3=10, Qe2=10";
6k1/1b3p1p/pq1r1np1/1p6/1N3Q2/5P1P/PPB3P1/3R3K b - - bm Ne8; id "Michael Hart.028"; c0 "Ne8=10, Rd8=10, Re6=8, Rxd1+=10";
2r2rk1/3R1pp1/7p/4p2P/4Q3/1Pq2P2/P7/1K1R4 b - - bm Kh8; id "Michael Hart.029"; c0 "Kh8=10, Rc5=5, Rc7=5, Rfe8=3";
2r1r1k1/pp2qppp/1n2b3/8/3PpNP1/1P2PB1P/P2Q4/2R1R1K1 w - - bm Bh1; id "Michael Hart.030"; c0 "Bh1=10, Bg2=10, Bxe4=8, Rxc8=10";
r1n2rk1/1p3ppp/p1n1b3/2P1p3/4P3/2P1BP2/1N1R2PP/R4BK1 b - - bm N8e7; id "Michael Hart.031"; c0 "N8e7=10, Rd8=4, Re8=5, h6=5";
1r2r1k1/1q1bbppp/5n2/1p2p3/p3P3/3BB2P/PP1NQPP1/2R1R1K1 w - - bm Nf3; id "Michael Hart.032"; c0 "Nf3=10, Bc5=10, Red1=8, a3=6";
8/3k4/2p3p1/p1Bnr3/Pp2p2P/1P4P1/2PK4/4R3 b - - bm Re8; id "Michael Hart.033"; c0 "Re8=10, Kc7=6, Nc3=5, Re6=6";
6k1/pq2ppb1/1p4pp/n7/3P3B/3QPN1P/P4PP1/6K1 w - - bm Qb5; id "Michael Hart.034"; c0 "Qb5=10, Qc3=4, a4=4, d5=8";
7r/3k2bp/p7/8/3p1B2/1Pp5/P1P2KPP/3R4 b - - bm h5; id "Michael Hart.035"; c0 "h5=10, Kc6=5, Re8=4, Rf8=2";
3rr1k1/1p2bppp/pn6/8/8/2P2N2/PP3PPP/R1B1R1K1 w - - bm Kf1; id "Michael Hart.036"; c0 "Kf1=10, Rb1=9, b3=10, h3=9";
3r3k/6b1/R1n2r1p/1q1p2pB/4p2P/2P3B1/5PP1/R1Q3K1 b - - bm Kh7; id "Michael Hart.037"; c0 "Kh7=10, Qc5=4, Rdf8=4, d4=3";
2r5/p4pk1/3Qbqp1/1p2p2p/2r1P2P/1NP2P2/PP4P1/1K1R3R w - - bm Qd2; id "Michael Hart.038"; c0 "Qd2=10, Ka1=10, Rd2=10, Rd3=9";
1r4k1/1bq1bppp/3p1n2/p1n1pP2/4P1P1/1P1B3Q/NP1B3P/1K1RR3 w - - bm Nc3; id "Michael Hart.039"; c0 "Nc3=10, Bc1=9, Rc1=9, g5=9";
2r3k1/p1pqrpp1/1pn2b1p/3P4/2P2B2/P2Q1N1P/5PP1/1R2R1K1 w - - bm Rbc1; id "Michael Hart.040"; c0 "Rbc1=10, Bh2=5, Rbd1=7, Rxe7=7";
3r1r2/1p3pkp/p1bRp3/4Pp1B/8/1NN5/P1P3PP/6K1 w - - bm Rxd8; id "Michael Hart.041"; c0 "Rxd8=10, Na5=5, Rd1=5, Rd3=5";
3rr1k1/pb3ppp/8/1p1n4/8/PB3N2/1P3PPP/3R1RK1 w - - bm h3; id "Michael Hart.042"; c0 "h3=10, Rd4=10, Rde1=10, Rfe1=10";
1r3rk1/2q3pp/3b1p2/p1pPp3/PpQ5/5P2/1PP1RBPP/R5K1 w - - bm Rc1; id "Michael Hart.043"; c0 "Rc1=10, Be1=9, Re3=9, b3=10";
r3r3/pp3pk1/2b2qp1/7p/5n1P/2N3Q1/PP3PP1/R2R1BK1 w - - bm a3; id "Michael Hart.044"; c0 "a3=10, Rac1=4, Rd2=5, f3=4";
3r2k1/1ppr1qpp/2n1b3/p2np3/8/PP1P1NP1/1BQ1PKBP/2RR4 w - - bm Rf1; id "Michael Hart.045"; c0 "Rf1=10, Qc5=2, Rb1=5, Re1=5";
bqr1r1k1/5pbp/pp1ppnp1/2n5/2P1P3/1PN2PP1/PBN1R1BP/3RQ2K w - - bm Qd2; id "Michael Hart.046"; c0 "Qd2=10, Ba1=10, Na3=7, Qf1=7";
r1k2b1r/2q2p2/p2p1P2/2nP3p/1p1N2b1/8/PPP2Q1P/1K2RBR1 w - - bm Nc6; id "Michael Hart.047"; c0 "Nc6=10, Qh4=5, h3=10";
1r4k1/1b2bppp/3p4/p4P2/N2RP3/4R3/1PB4n/1K6 b - - bm Ng4; id "Michael Hart.048"; c0 "Ng4=10, Ba6=10, Ba8=2, Bc6=10";
5qk1/1prn2pp/p1r1pb2/4n3/3QPB2/P1N3PP/1PP5/1K1R1B1R w - - bm Be2; id "Michael Hart.049"; c0 "Be2=10, Bd3=6, Qd2=5";
2r2rk1/1n1qb1pp/pn6/2pPp3/p3N3/2P1BN1P/5PP1/R2QR1K1 w - - bm Qb1; id "Michael Hart.050"; c0 "Qb1=10, Bg5=10, Nxe5=10, Rb1=10";
r3r1k1/3p1ppp/2b2n2/4n1B1/Pq2P3/2N3PP/3Q2B1/4RRK1 w - - bm Rf5; id "Michael Hart.051"; c0 "Rf5=10, Bxf6=10, Kh2=1, Ra1=1";
5rk1/1p1qb2p/p2p1pp1/P2Pn3/1PrN4/1Q6/6PP/1R3RBK b - - bm Rfc8; id "Michael Hart.052"; c0 "Rfc8=10, Qc8=10, Qg4=10, f5=10";
2rqrnk1/3bbpp1/p6p/3p4/1n1P1PP1/2N5/1P2N1QP/RBB3RK w - - bm g5; id "Michael Hart.053"; c0 "g5=10, Bd2=7, Qf3=7, Rd1=7";
2r2rk1/pq1n1pp1/4pn1p/1pp5/3P3B/P3PP2/1P2Q1PP/1N1R1RK1 w - - bm Nc3; id "Michael Hart.054"; c0 "Nc3=10, Bg3=8, dxc5=10, e4=6";
r4bk1/5pp1/p1b1qn1p/2p1B3/Pp1P4/4N3/1PB2PPP/R2Q2K1 w - - bm h3; id "Michael Hart.055"; c0 "h3=10, Bb3=10, Qe2=10, Rc1=10";
5r1k/pp2pq1p/2pr1np1/5P2/3P3P/P3P1QB/1P5K/2R3R1 b - - bm Qg7; id "Michael Hart.056"; c0 "Qg7=10, Ne8=7, Nh5=7, gxf5=5";
2q3k1/p1p2p1p/2p3p1/8/4R3/1PQ1P2P/P4PPK/3r4 b - - bm a6; id "Michael Hart.057"; c0 "a6=10, Qd7=5, Rd6=4, Rd7=5";
3r1r2/6kp/8/1R3p2/3b1p2/8/3N1PPP/5RK1 b - - bm Rd6; id "Michael Hart.058"; c0 "Rd6=10, Bc3=5, Kf6=5, Rb8=1";
3r2k1/5pp1/2n1p2p/2p1P3/n3NP2/PpBr4/1P2RKPP/2R5 b - - bm R8d5; id "Michael Hart.059"; c0 "R8d5=10, Kh7=9, c4=10, h5=9";
4qr1k/3b2pp/p1rpp3/1p5n/4P1B1/2N1QP2/PPP4P/1K1R2R1 w - - bm e5; id "Michael Hart.060"; c0 "e5=10, Bxh5=6, a3=6, h4=6";

Jouni
Posts: 2393
Joined: Wed Mar 08, 2006 7:15 pm

Re: new epd test suite

Post by Jouni » Tue Oct 20, 2009 7:56 pm

Interesting stuff. I have tried quite similar many times, but finally given up. There are always engines, which seems ruin every test suite. One notable engine is Comet B.68: in this suite that 2400 rated engine is faster than 2800 rated bright 0.4a!

Jouni

Dann Corbit
Posts: 12040
Joined: Wed Mar 08, 2006 7:57 pm
Location: Redmond, WA USA
Contact:

Re: new epd test suite

Post by Dann Corbit » Tue Oct 20, 2009 8:16 pm

Jouni wrote:Interesting stuff. I have tried quite similar many times, but finally given up. There are always engines, which seems ruin every test suite. One notable engine is Comet B.68: in this suite that 2400 rated engine is faster than 2800 rated bright 0.4a!

Jouni
I can tune an engine {with an automated procedure} so that it easily aces a given test suite, but then it will play games a lot worse.

Good tactical ability and good strategic ability are sometimes at odds. If you tune for tactics, then your strategical strength can be reduced (depending on how you do the tuning).

Hart

Re: new epd test suite

Post by Hart » Wed Oct 21, 2009 12:00 am

Forget what I said about suggested usage, but note that this test is very sensitive to hardware and the engines meant to be tested and their strengths relative to what they're tuned for. For example, tuning parameters normally requires hundreds if not thousands of games over long periods of time. However, in this suite, in the right context, you really can see noticeable results in a short period of time. For instance, using only 20 seconds on 1 core, both Naum 4 with pruning at +10 and -10 do markedly worse than default on my latest (yet to be released) 60 positions at 20 seconds per, and I wouldn't call this a drastic change to the engine, but I may be wrong on that last point. Also, these are all middle game positions, but when I change SF's parameter for Passed Pawns (End Game), I also get markedly better results, when you (I) would generally not expect to see any noticeable difference at all. I compiled I***** with no optimizations and then did a general optimization with /Ox and I also detected a playing strength increase, though my optimization was far from being very optimized. I also decrease SF's passed pawn parameter (end game) from 100 to 95, and that is also detectable, and it has been my opinion from first seeing this engine play that it overvalued passes pawns in the endgame. Maybe this is normal, but it is far from what I would expect. My final test is going to be to run SF through these tests and them optimize the parameters accordingly and see if I am on to something, if not, then you can put a fork in me.

Hart

Re: new epd test suite

Post by Hart » Wed Oct 21, 2009 1:53 am

Code: Select all

-----------------Stockfish_151_ja-----------------
Stockfish_151_ja - Stockfish_151_ja MW      &#58; 2.5/10 1-6-3 (=0000010==                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                      )  25%  -191
-----------------Stockfish_151_ja MW-----------------
Stockfish_151_ja MW - Stockfish_151_ja      &#58; 7.5/10 6-1-3 (=1111101==
These are my results so far. I have only touched 4 or 5 parameters but the results are better than I would have expected for only 10 games. 6 wins and 1 loss. This is at very fast play, 1-core and 32bit versions using HS440 with alternating colors. Of course, we can only take 10 games so seriously but I will keep you posted.

Dann Corbit
Posts: 12040
Joined: Wed Mar 08, 2006 7:57 pm
Location: Redmond, WA USA
Contact:

Re: new epd test suite

Post by Dann Corbit » Wed Oct 21, 2009 6:23 pm

Rybka 4 thread {not multi-pv} analysis:

Code: Select all

Analysis from C&#58;\test\MichaelHart.epd   
Analyzing engine&#58; Rybka 3
10/20/2009 11&#58;34&#58;20 PM Level&#58; 660 Seconds
  1&#41; .. Bb3-d5          Bb3-d5  * 0 Seconds
  2&#41; .. Qe6-d5          Qe6-d5  * 0 Seconds
  3&#41; h4-h5              Bf1-b5   
  4&#41; h2-h4              h2-h4   * 21 Seconds
  5&#41; .. Rf8-b8          Rf8-b8  * 0 Seconds
  6&#41; .. h5-h4           h5-h4   * 0 Seconds
  7&#41; Nf3xe5             Nf3xe5  * 195 Seconds
  8&#41; .. Qe6-f5          Qe6-f5  * 0 Seconds
  9&#41; a3-a4              a3-a4   * 6 Seconds
 10&#41; .. h7-h6           h7-h6   * 2 Seconds
 11&#41; .. Rc6-c5          Rc6-c5  * 0 Seconds
 12&#41; Qd3-e4             Qd3-e4  * 0 Seconds
 13&#41; Ne2-c1             Ne2-c1  * 124 Seconds
 14&#41; .. Qd8-b6          Qd8-d6   
 15&#41; Qg3-d6             Qg3-d6  * 0 Seconds
 16&#41; Rd1xd4             Rd1xd4  * 0 Seconds
 17&#41; .. Nc6-d4          Nc6-e7   
 18&#41; f2-f4              f2-f4   * 0 Seconds
 19&#41; .. Nd5-f6          Nd5-f6  * 157 Seconds
 20&#41; .. Kf8xe8          Kf8xe8  * 0 Seconds
 21&#41; Nb5-c7             Nb5-c7  * 0 Seconds
 22&#41; Na4-c3             Na4-c3  * 3 Seconds
 23&#41; Qd2xb4             Qd2xb4  * 0 Seconds
 24&#41; .. Qf6-f5          Qf6-f5  * 0 Seconds
 25&#41; Bg5-h4             Bg5-h4  * 0 Seconds
 26&#41; b2xc3              b2xc3   * 0 Seconds
 27&#41; Qd2-e2             Qd2-e2  * 386 Seconds
 28&#41; .. Rd6xd1          Rd6xd1  * 0 Seconds
 29&#41; .. Rf8-e8          Kg8-h8   
 30&#41; Rc1xc8             Rc1xc8  * 0 Seconds
 31&#41; .. Rf8-d8          Rf8-d8  * 0 Seconds
 32&#41; Nd2-f3             Nd2-f3  * 13 Seconds
 33&#41; .. Re5-e8          Re5-e8  * 0 Seconds
 34&#41; Qd3-b5             Qd3-b5  * 10 Seconds
 35&#41; .. Rh8-f8          Rh8-f8  * 0 Seconds
 36&#41; b2-b3              b2-b3   * 1 Second
 37&#41; .. d5-d4           d5-d4   * 0 Seconds
 38&#41; Rd1-d2             Rd1-d3   
 39&#41; Na2-c3             Na2-c3  * 0 Seconds
 40&#41; Rb1-d1             Rb1-d1  * 48 Seconds
 41&#41; Rd6-d1             Nb3-a5   
 42&#41; Rf1-e1             Rd1-d4   
 43&#41; b2-b3              b2-b3   * 27 Seconds
 44&#41; f2-f3              f2-f3   * 0 Seconds
 45&#41; Qc2-c5             Qc2-c5  * 0 Seconds
 46&#41; Qe1-d2             Bb2-a1   
 47&#41; h2-h3              h2-h3   * 0 Seconds
 48&#41; .. Bb7-c6          Bb7-c6  * 1 Second
 49&#41; Qd4-b4             Qd4-b4  * 22 Seconds
 50&#41; Ra1-b1             Ra1-b1  * 26 Seconds
 51&#41; Bg5xf6             Bg5xf6  * 0 Seconds
 52&#41; .. Qd7-g4          Qd7-g4  * 87 Seconds
 53&#41; g4-g5              g4-g5   * 0 Seconds
 54&#41; Nb1-c3             d4xc5    
 55&#41; Ra1-c1             Ra1-c1  * 0 Seconds
 56&#41; .. Qf7-g7          g6xf5    
 57&#41; .. Rd1-d6          Rd1-d6  * 1 Second
 58&#41; .. Rd8-b8          Rd8-b8  * 0 Seconds
 59&#41; .. c5-c4           Kg8-h7   
 60&#41; e4-e5              e4-e5   * 0 Seconds
49 of 60 matching moves
10/21/2009 10&#58;35&#58;15 AM, Total time&#58; 11&#58;00&#58;55 AM Rated time&#58; 2&#58;19&#58;50 = 8390 Seconds

Dann Corbit
Posts: 12040
Joined: Wed Mar 08, 2006 7:57 pm
Location: Redmond, WA USA
Contact:

Re: new epd test suite problem 3

Post by Dann Corbit » Wed Oct 21, 2009 6:25 pm

[d]rr4k1/2q1bppp/3p4/p2Rn1P1/1p1NPQ1P/8/PPP5/1K3B1R w - - fmvn 22; hmvc 2; bm h5; id "Michael Hart.003";

Code: Select all

3&#41; h5; id "Michael Hart 
    Searching move&#58; h4-h5
    Best move &#40;Rybka 3&#41;&#58; Bf1-b5
    Not found in&#58; 11&#58;00
      2	00&#58;00	         450	13.963	+0.37	h4h5
      3	00&#58;00	         877	27.213	+0.37	h4h5
      4	00&#58;00	       1.685	52.286	+0.37	h4h5
      4	00&#58;00	       2.055	63.767	+0.38	Qf4d2
      5	00&#58;00	       3.656	77.994	+0.51	Qf4d2 Be7f8
      6	00&#58;00	       7.090	113.440	+0.53	Qf4d2 Be7f8 h4h5
      7	00&#58;00	      18.559	150.828	+0.49	Qf4d2 a5a4 Nd4b5 Qc7c6 Qd2xb4
      8	00&#58;00	      31.842	172.519	+0.51	Qf4d2 a5a4 Nd4b5 Qc7b6 Qd2xb4 Qb6e3
      9	00&#58;00	      48.281	185.167	+0.38	Qf4d2 a5a4 Nd4b5 Qc7c8 Qd2xb4 Qc8g4 Bf1d3 Qg4f3
     10	00&#58;00	      96.394	185.540	+0.34	Qf4d2 a5a4 Nd4b5 Qc7c8 Qd2xb4 Qc8g4 Bf1d3 Ne5xd3 Rd5xd3 a4a3
     10	00&#58;00	     128.475	187.138	+0.37	Nd4b5 Qc7c6 Qf4d2 Ne5f3 Qd2e3 Nf3e5 Nb5d4
     11	00&#58;00	     140.324	187.587	+0.34	Nd4b5 Qc7c6 Qf4d2 Ne5f3 Qd2e3 Nf3e5 Nb5d4 Qc6c7 Bf1e2
     12	00&#58;01	     218.763	196.502	+0.34	Nd4b5 Qc7c6 Qf4d2 Ne5f3 Qd2e3 Nf3e5 Nb5d4 Qc6c7 Bf1e2
     13	00&#58;02	     364.946	208.076	+0.34	Nd4b5 Qc7c6 Qf4d2 Ne5f3 Qd2e3 Nf3e5 Nb5d4 Qc6c7 Bf1e2
     13	00&#58;08	   1.721.440	213.563	+0.48	h4h5 b4b3 c2xb3 a5a4 h5h6
     14	00&#58;11	   2.352.352	215.322	+0.29	h4h5 b4b3 c2xb3 a5a4 a2a3 a4xb3 Bf1b5 Rb8c8
     14	00&#58;16	   3.298.448	207.968	+0.35	Nd4b5 Qc7c6 Qf4d2 Ne5f3 Qd2d1 Nf3e5 Rh1h3 Rb8c8 Nb5d4 Qc6c7 Qd1d2
     14	00&#58;22	   4.467.855	201.963	+0.56	Qf4d2 a5a4 Bf1b5 Ne5c4 Bb5xc4 Qc7xc4 Qd2d3 Qc4c8 Rd5f5 b4b3 c2xb3 Qc8e8 Rh1f1 a4xb3
     15	00&#58;46	   9.291.214	203.695	+0.58	Qf4d2 Qc7b7 Bf1b5 Rb8c8 Rh1d1 Be7f8 Bb5a4 Rc8c5 Nd4f5 Rc5xd5 Qd2xd5 Qb7xd5 Rd1xd5 Ra8a6 Ba4b5 Ra6b6
     15	00&#58;49	   9.670.579	202.091	+0.69	Bf1b5 a5a4 Qf4d2 b4b3 c2xb3 a4xb3 Nd4xb3 Rb8c8 Rh1c1 Qc7d8 Rc1xc8 Qd8xc8 Qd2g2 Qc8c7 a2a4 Ra8c8 a4a5 Qc7a7 a5a6 Qa7e3
     16	01&#58;04	  12.162.856	194.626	+0.71	Bf1b5 a5a4 Qf4d2 Ne5c4 Bb5xc4 Qc7xc4 Qd2d3 Qc4c7 Nd4b5 Qc7b6 Rh1f1 Ra8a5 Nb5d4 Ra5xd5 e4xd5 a4a3 Nd4c6 Rb8b7 b2b3 g7g6 Qd3e2 Be7f8
     17	01&#58;17	  14.745.217	195.134	+0.71	Bf1b5 a5a4 Qf4d2 Ne5c4 Bb5xc4 Qc7xc4 Qd2d3 Qc4c7 Nd4b5 Qc7b6 Rh1f1 Ra8a5 Nb5d4 Ra5xd5 e4xd5 a4a3 Nd4c6 Rb8b7 b2b3 g7g6 Qd3e2 Be7f8
     18	02&#58;59	  33.994.126	194.167	+0.77	Bf1b5 a5a4 Qf4d2 Ne5c4 Bb5xc4 Qc7xc4 Qd2d3 Rb8c8 Nd4f5 Be7f8 Nf5xd6 Qc4xd3 c2xd3 Rc8d8
     19	04&#58;17	  50.046.646	199.566	+0.77	Bf1b5 a5a4 Qf4d2 Ne5c4 Bb5xc4 Qc7xc4 Qd2d3 Rb8c8 Nd4f5 Be7f8 Nf5xd6 Qc4xd3 c2xd3 Rc8d8
     20	06&#58;29	  76.620.209	201.465	+0.73	Bf1b5 a5a4 Qf4d2 Ne5c4 Bb5xc4 Qc7xc4 Qd2d3 Rb8c8 Nd4f5 Be7f8 Nf5xd6 Qc4xd3 c2xd3 Rc8d8 e4e5 Ra8a6 Rh1c1 b4b3 d3d4 Bf8xd6 Rd5xd6 Ra6xd6 e5xd6 f7f5 g5xf6/ep g7xf6 Rc1c4 b3xa2+ Kb1xa2 Rd8xd6 Rc4xa4
   10/21/2009 12&#58;07&#58;34 AM, Time for this analysis&#58; 00&#58;11&#58;00, Rated time&#58; 11&#58;00

Dann Corbit
Posts: 12040
Joined: Wed Mar 08, 2006 7:57 pm
Location: Redmond, WA USA
Contact:

Re: new epd test suite problem 14

Post by Dann Corbit » Wed Oct 21, 2009 6:26 pm

[d]3qr1k1/pp6/6b1/3PpnB1/2r5/2N4P/PP1Q2B1/3R3K b - - fmvn 30; hmvc 3; bm Qb6; id "Michael Hart.014";

Code: Select all

14&#41; Qb6; id "Michael Ha 
    Searching move&#58; Qd8-b6
    Best move &#40;Rybka 3&#41;&#58; Qd8-d6
    Not found in&#58; 11&#58;00
      2	00&#58;00	         416	425.984	+0.42	Qd8d6
      3	00&#58;00	         802	821.248	+0.37	Qd8d6
      3	00&#58;00	       1.055	1.080.320	+0.41	Qd8d7
      4	00&#58;00	       1.602	96.496	+0.24	Qd8d7
      4	00&#58;00	       2.492	150.106	+0.35	Qd8d6
      5	00&#58;00	       5.308	169.856	+0.39	Qd8d6 Qd2f2
      6	00&#58;00	       6.313	202.016	+0.33	Qd8d6 Qd2f2 Nf5d4
      7	00&#58;00	      14.360	186.134	+0.25	Qd8d6 Qd2f2 Rc4d4 Rd1e1
      7	00&#58;00	      27.494	199.672	+0.49	Qd8b6 Rd1g1 Re8f8 Kh1h2
      8	00&#58;00	      47.991	195.787	+0.31	Qd8b6 Rd1g1 Re8f8 d5d6 Rc4d4 Qd2e1 Rd4d3 Qe1xe5
      9	00&#58;00	      57.878	189.351	+0.31	Qd8b6 Rd1g1 Re8f8 d5d6 Rc4d4 Qd2e1 Rd4d3 Qe1xe5
     10	00&#58;00	      82.577	186.664	+0.31	Qd8b6 Rd1g1 Re8f8 d5d6 Rc4d4 Qd2e1 Rd4d3 Qe1xe5
     11	00&#58;01	     172.268	191.533	+0.29	Qd8b6 Rd1g1 Re8f8 d5d6 Rc4d4 Qd2e1 Rd4d3 Qe1xe5 Nf5g3+ Kh1h2
     12	00&#58;02	     423.192	198.328	+0.30	Qd8b6 Rd1g1 Re8f8 d5d6 Rc4d4 Qd2e1 Rd4xd6 Qe1xe5 Rf8e8 Qe5h2 Kg8h7 Bg2e4
     13	00&#58;07	   1.347.629	205.200	+0.32	Qd8b6 Rd1g1 Rc4d4 Qd2f2 e5e4 Bg2f1 e4e3 Qf2f3 Rd4d2 Qf3xf5 Bg6xf5 Bg5xe3+ Kg8f7 Be3xb6 a7xb6 Rg1g2 Rd2xg2 Kh1xg2 Kf7f6 h3h4 Re8g8+ Kg2f3
     14	00&#58;09	   1.913.740	207.614	+0.31	Qd8b6 Rd1g1 Rc4d4 Qd2f2 e5e4 Rg1e1 e4e3 Bg5xe3 Nf5xe3 Re1xe3 Re8xe3 Qf2xe3 a7a6 Qe3g5 Kg8f7 Qg5g3 Rd4d3
     15	00&#58;25	   5.384.450	215.765	+0.31	Qd8b6 Rd1g1 Rc4d4 Qd2f2 e5e4 Bg2f1 e4e3 Qf2f3 Rd4d2 Qf3xf5 Bg6xf5 Bg5xe3+ Kg8f7 Be3xb6 a7xb6 Rg1g2 Rd2xg2 Kh1xg2 Re8g8+ Kg2f3 Kf7f6 Kf3f4 Rg8h8
     16	00&#58;49	  10.137.110	210.568	+0.28	Qd8b6 Rd1g1 Rc4d4 Qd2f2 e5e4 Bg2f1 e4e3 Qf2f3 Rd4d2 Qf3xf5 Bg6xf5 Bg5xe3+ Kg8f7 Be3xb6 a7xb6 Rg1g2 Rd2xg2 Kh1xg2 Re8g8+ Kg2f2 Kf7e7 h3h4 Rg8h8 Kf2g3 Ke7d6 Bf1e2 Bf5g6
     17	05&#58;41	  73.507.497	220.302	+0.21	Qd8d6 Nc3b5 Qd6d7 d5d6 Kg8h7 Qd2d5 Rc4c6 Qd5f3 Kh7g7 Rd1e1 a7a6 Nb5c7 Re8c8 Re1xe5 Rc6xd6 Nc7d5 Rd6e6 Qf3f4 Rc8e8
     18	08&#58;36	 111.240.802	220.874	+0.18	Qd8d6 Nc3b5 Qd6d7 d5d6 Kg8h7 Qd2d5 Rc4b4 a2a4 a7a6 Rd1c1 Bg6f7 Qd5xb7 Qd7xb7 Bg2xb7 a6xb5 d6d7 Re8g8 Rc1c8 Rb4d4 d7d8Q Rd4xd8 Bg5xd8 Nf5d6 Rc8b8 Nd6xb7 Rb8xb7 Kh7g6
   10/21/2009 2&#58;08&#58;42 AM, Time for this analysis&#58; 00&#58;11&#58;00, Rated time&#58; 27&#58;48

Post Reply