Is Komodo 4 too aggressive or is it contempt?
Evaluation of Komodo 4
-
kingliveson
- Posts: 1388
- Joined: Thu Jun 10, 2010 1:22 am
- Real Name: Franklin Titus
- Location: 28°32'1"N 81°22'33"W
Evaluation of Komodo 4
PAWN : Knight >> Bishop >> Rook >>Queen
-
kingliveson
- Posts: 1388
- Joined: Thu Jun 10, 2010 1:22 am
- Real Name: Franklin Titus
- Location: 28°32'1"N 81°22'33"W
Re: Evaluation of Komodo 4
I wanted to see how Komodo 4 evaluates and performs with increased time control -- this time, 60 minutes, ponder on, no tablebases, no openings, default settings, 512 MB hash, and single core (AMD Phenom II x4 @ 3000 MHz).
The first 11 rounds:
Round 12 was quite an interesting match, and definitely one for the books. Rook sacrifice for Passed pawn proved a worthy attack strategy from beginning to end:
The first 11 rounds:
Round 12 was quite an interesting match, and definitely one for the books. Rook sacrifice for Passed pawn proved a worthy attack strategy from beginning to end:
PAWN : Knight >> Bishop >> Rook >>Queen
-
kingliveson
- Posts: 1388
- Joined: Thu Jun 10, 2010 1:22 am
- Real Name: Franklin Titus
- Location: 28°32'1"N 81°22'33"W
Re: Evaluation of Komodo 4
Code: Select all
xboard -fcp Komodo -scp Pawno -fUCI -sUCI -defaultHashSize 128 -tc 3 -inc 0 -mg 50Code: Select all
xboard: Match Komodo64 SSE Version 4 vs. Pawno 1.0b x64: final score 8-20-22- Attachments
-
- 20120211_komodo_v_pawno_3m_50r.pgn
- Ponder on, single-core, 25 positions
- (200.4 KiB) Downloaded 308 times
PAWN : Knight >> Bishop >> Rook >>Queen
Re: Evaluation of Komodo 4
I'll bite....... any info on Pawno 1.0b ?
-
kingliveson
- Posts: 1388
- Joined: Thu Jun 10, 2010 1:22 am
- Real Name: Franklin Titus
- Location: 28°32'1"N 81°22'33"W
-
kingliveson
- Posts: 1388
- Joined: Thu Jun 10, 2010 1:22 am
- Real Name: Franklin Titus
- Location: 28°32'1"N 81°22'33"W
Re: Evaluation of Komodo 4
Komodo 4 vs Houdini 1.5a (wine). 1-cpu (AMD Phenom @ 3000 MHz), Hash 512 MB, 60 minutes. Note that Houdini and Critter 1.4 (Linux) combined with Xboard/PolyGlot only use half the Hash allotted. The match ended abruptly due to power going out for a split second.
The 1st 15 rounds:
The last 14 rounds:
The 1st 15 rounds:
The last 14 rounds:
PAWN : Knight >> Bishop >> Rook >>Queen
Re: Evaluation of Komodo 4
I haven't looked at them all yet, but Houdini's wins in games 23 and 27 (in 170+ and 210+ moves) are telling.Is Komodo 4 too aggressive or is it contempt?
Re: Evaluation of Komodo 4
Here are brief comments from Games 1-15:
Game 1: A long game, in which Komodo won an exchange for a pawn, but it didn't lead to anything (QRR+3 to QRB+4).
Game 2: Houdini reached +2.39 in QR+5 (with two connected passers) vs RRBN+2, but this rather radpily dwindled to nothing.
Game 3: Houdini gained a lot moves 25-30 by increasing Q-side pressure.
Game 4: Komodo got into a RB+5 vs Q+4 endgame, and won it with the passed c-pawn. Queens are not good at blocking passers.
Game 5: A rapid turnabout around move 27-32, with Komodo getting 2 pawns for a piece. I don't think Black really had an edge, with the passed h-pawn eventually just winning the piece back.
Game 6: Again nothing much happens for 50 moves, and then Komodo wins an endgame via a passed a-pawn (joined by a c-pawn).
Game 7: Houdini speculates with 20. Nxb7 and loses the ensuing QR vs RBBN. Three pieces often beat a queen when they coordinate and there are no targets.
Game 8: Houdini's turn to win with a passed pawn, here in a RBN vs RR endgame.
Game 9: Komodo got a rather good early edge with Black here, sacrificed a piece with 19. Bxh4, and powered the attack home in the end.
Game 10: Somewhat of a reverse of Game 1. Houdini had QRR vs QRB, and managed to get a passed pawn after exchanging queens, but nothing came of it.
Game 11: Houdini again won via a Q-side attack. Komodo likes to O-O-O behind this Exchange Ruy pawn structure, and I'm not sure how great that really is in many cases (see previous games 3, 7, and maybe 9).
Game 12: Houdini eventually got a minor edge from Q+3 vs RR+2, but there was no real progress (even after White won a pawn).
Game 13: See Game 11, though the breakthrough was rather different this time.
Game 14: Rather boring draw, swapping down to RNN at move 25.
Game 15: Houdini reached +2.92 in N+3 vs B+2, but failed to convert. White had passed c- and h-pawns, but knights do not help these much.
Game 1: A long game, in which Komodo won an exchange for a pawn, but it didn't lead to anything (QRR+3 to QRB+4).
Game 2: Houdini reached +2.39 in QR+5 (with two connected passers) vs RRBN+2, but this rather radpily dwindled to nothing.
Game 3: Houdini gained a lot moves 25-30 by increasing Q-side pressure.
Game 4: Komodo got into a RB+5 vs Q+4 endgame, and won it with the passed c-pawn. Queens are not good at blocking passers.
Game 5: A rapid turnabout around move 27-32, with Komodo getting 2 pawns for a piece. I don't think Black really had an edge, with the passed h-pawn eventually just winning the piece back.
Game 6: Again nothing much happens for 50 moves, and then Komodo wins an endgame via a passed a-pawn (joined by a c-pawn).
Game 7: Houdini speculates with 20. Nxb7 and loses the ensuing QR vs RBBN. Three pieces often beat a queen when they coordinate and there are no targets.
Game 8: Houdini's turn to win with a passed pawn, here in a RBN vs RR endgame.
Game 9: Komodo got a rather good early edge with Black here, sacrificed a piece with 19. Bxh4, and powered the attack home in the end.
Game 10: Somewhat of a reverse of Game 1. Houdini had QRR vs QRB, and managed to get a passed pawn after exchanging queens, but nothing came of it.
Game 11: Houdini again won via a Q-side attack. Komodo likes to O-O-O behind this Exchange Ruy pawn structure, and I'm not sure how great that really is in many cases (see previous games 3, 7, and maybe 9).
Game 12: Houdini eventually got a minor edge from Q+3 vs RR+2, but there was no real progress (even after White won a pawn).
Game 13: See Game 11, though the breakthrough was rather different this time.
Game 14: Rather boring draw, swapping down to RNN at move 25.
Game 15: Houdini reached +2.92 in N+3 vs B+2, but failed to convert. White had passed c- and h-pawns, but knights do not help these much.
Re: Evaluation of Komodo 4
Game 16: Both sides shuffled around a bit, with doubled rooks on half-open files, then the fireworks game, and Black ended up with a passed d-pawn (joined by a passed c-pawn, on the 6th rank) in return for an exchange. Eventually they went into a drawn queen ending.
Game 17: Komodo speculated with 39. Nxf4, and eventually lost. Again we saw Black O-O-O behind the Exchange Ruy dxc6, but here neither side was able to attack the opposing King too readily via a pawn storm.
Game 18: I didn't see much in this. Except for both engines overestimating White's chances in the R+2 vs R+1 endgame (and the ensuing K+2 vs K+1).
Game 19: White never got a pawn storm against Black's O-O-O Exchange Ruy setup, and Black was able to win a bunch of space, followed by a slow penetration and victory.
Game 20: Again looked pretty drawish. No real open lines, and the Black Nd5 was immovable. Komodo was claiming a reasonable edge at some points, but I didn't really see it.
Game 21: Both sides agreed White had the edge until about move 30, at which point Black's King proved safe, while White's h4 pawn was weak, and Black (Komodo) won after beating off White's Q-side pressure.
Game 22: Somewhat like Game 16, but w/o the queens. Eventually Houdini got saddled with a lousy bishop, and White had holes for the knights, and Black suffered.
Game 23: Classic board-block. Komodo shuffled around for 50 moves, then some more, finally pushed g4 (move 114), at which point Houdini agreed Black had the advantage, but 15 moves later it was White who had the edge, eventually converted. I don't whether the 2 bishops are overvalued in this sort of position, or what.
Game 24: Both sides had activity, but neither seemed threatening to win.
Game 25: Again a battle of Q-side pressure against O-O-O behind dxc6 versus pushing K-side pawns. Houdini prevailed with c6 (giving up a piece) and then b5, with White's Queen being two agile against RBB in the endgame.
Game 26: Komodo seemed to allow Black to expand on both sides, and eventually Houdini penetrated via the h-file.
Game 27: A lot like Game 23, though here it was White (Houdini) who made the break with b4. After almost 100 moves of shuffling in a QN vs QB with blocked pawns, White's King entered the picture, winning the f-pawn and
the game.
Game 28: Black had an advantage here, but the passed a-pawn had no real support in the end.
Game 29: Komodo got in f5/f4 fairly early, and while Houdini pushed b4/b5 a few moves later, White's King was more in trouble. After winning a couple of pawns, Komodo exchanged down and won the endgame.
Game 17: Komodo speculated with 39. Nxf4, and eventually lost. Again we saw Black O-O-O behind the Exchange Ruy dxc6, but here neither side was able to attack the opposing King too readily via a pawn storm.
Game 18: I didn't see much in this. Except for both engines overestimating White's chances in the R+2 vs R+1 endgame (and the ensuing K+2 vs K+1).
Game 19: White never got a pawn storm against Black's O-O-O Exchange Ruy setup, and Black was able to win a bunch of space, followed by a slow penetration and victory.
Game 20: Again looked pretty drawish. No real open lines, and the Black Nd5 was immovable. Komodo was claiming a reasonable edge at some points, but I didn't really see it.
Game 21: Both sides agreed White had the edge until about move 30, at which point Black's King proved safe, while White's h4 pawn was weak, and Black (Komodo) won after beating off White's Q-side pressure.
Game 22: Somewhat like Game 16, but w/o the queens. Eventually Houdini got saddled with a lousy bishop, and White had holes for the knights, and Black suffered.
Game 23: Classic board-block. Komodo shuffled around for 50 moves, then some more, finally pushed g4 (move 114), at which point Houdini agreed Black had the advantage, but 15 moves later it was White who had the edge, eventually converted. I don't whether the 2 bishops are overvalued in this sort of position, or what.
Game 24: Both sides had activity, but neither seemed threatening to win.
Game 25: Again a battle of Q-side pressure against O-O-O behind dxc6 versus pushing K-side pawns. Houdini prevailed with c6 (giving up a piece) and then b5, with White's Queen being two agile against RBB in the endgame.
Game 26: Komodo seemed to allow Black to expand on both sides, and eventually Houdini penetrated via the h-file.
Game 27: A lot like Game 23, though here it was White (Houdini) who made the break with b4. After almost 100 moves of shuffling in a QN vs QB with blocked pawns, White's King entered the picture, winning the f-pawn and
the game.
Game 28: Black had an advantage here, but the passed a-pawn had no real support in the end.
Game 29: Komodo got in f5/f4 fairly early, and while Houdini pushed b4/b5 a few moves later, White's King was more in trouble. After winning a couple of pawns, Komodo exchanged down and won the endgame.
-
kingliveson
- Posts: 1388
- Joined: Thu Jun 10, 2010 1:22 am
- Real Name: Franklin Titus
- Location: 28°32'1"N 81°22'33"W
Re: Evaluation of Komodo 4
I think I read somewhere, either Rybka Forum or Talkchess, where the authors wrote that Komodo 4 is definitely more aggressive compared to previous versions.BB+ wrote:I haven't looked at them all yet, but Houdini's wins in games 23 and 27 (in 170+ and 210+ moves) are telling.Is Komodo 4 too aggressive or is it contempt?
Maybe this is one of the side effects of high queen value...Game 4 too.Game 7: Houdini speculates with 20. Nxb7 and loses the ensuing QR vs RBBN. Three pieces often beat a queen when they coordinate and there are no targets.
PAWN : Knight >> Bishop >> Rook >>Queen