Komodo Dragon 2 released.

[Chessqueen]

Komodo Dragon 2 is released today at komodochess.com. As usual, it is free for subscribers, priced the same as Dragon (1), which itself will be marked down 20% as will Komoro 14.1 shortly. Anyone who bought Komodo 14 or later version gets 20% discount from these prices.
The improvements over Dragon 1 are primarily in two areas. MultiPV search (in standard mode) has been completely rewritten, and now is
about a hundred elo stronger than in Dragon 1 (or older Komodo versions). It is sufficiently improved that we no longer advocate using MCTS mode
just for the MultiPV benefits it provides. The MCTS mode is also improved in Dragon 2, dramatically so when using Windows, where the elo gain
over the original Dragon is also about a hundred elo. There is a new UCI option called MCTS Optimism, which when set to a positive number
(default is +10) makes Dragon "trust" its own internal move preferences more while exploring the opponent's choices more; a higher setting
might be appropriate for a human opponent. MCTS mode remains the best option for super-strong but human-like play and for preparing openings for
human opponents who will not have access to engines while playing. The standard mode when not using MultiPV has also been improved, but only by
an estimated ten elo. It is nearly the same as the version that won the recent TCEC Swiss, and is a slight upgrade from the version that beat
Stockfish in the fifty game final match of the TCEC chess960 (FRC) championship.
We believe that the MultiPV search is now more effective than the Stockfish MultiPV search, and that if MultiPV is set to more than 4
Dragon 2 is now the favorite against Stockfish 13 on typical modern hardware.
Dragon 2 comes with an AVX2 version (suitable for most PCs since 2013) and a standard one. The AVX2 version is quite a bit faster and
stronger, and we hope that the testing groups will limit testing to AVX2 machines if at all possible. It probably doesn't matter when running
against Stockfish 12 or later, which also gets a lot out of AVX2, but it makes a big difference against programs not using NNUE. There is no
BMI2 version for Dragon 2 as we have not found any meaningful benefit to BMI2 beyond the benefit of AVX2.
Finally, for those who bought Komodo 14 or 14.1 but not Dragon, the strength gains from Komodo 14 in either MCTS or standard mode are nearly
200 elo, only slightly less from Komodo 14.1. All estimated elo gains are at the CCRL blitz time control of 2' + 1" increment, and use Ordo rather
than BayesElo.

I am just curious, is a 2 pawns odds plus Non Castling for Dragon 2 Vs GM Nakamura worse or better than Knight odds at 2'+1" T/C

[carldaman]

Dragon 2 is now the favorite against Stockfish 13 on typical modern hardware.


Rgds

Is it? They seem to imply that with a pv of one the improvement is small, but the difference between Dragon 1 and SF 13 is 79 elo:

http://www.cegt.net/40_40%20Rating%20Li ... liste.html

The above is a fragment of the whole sentence. Larry wrote "We believe that the MultiPV search is now more effective than the Stockfish MultiPV search, and that if MultiPV is set to more than 4 Dragon 2 is now the favorite against Stockfish 13 on typical modern hardware."

So if you set both Dragon 2 and Stockfish to use MultiPV of 4 or more, Dragon 2 will be stronger in games, which basically means you will get better analysis quicker with Dragon 2 displaying 4 or more lines. Basically we improved MultiPV a lot in Dragon 2 (and frankly the old MultiPV scheme was not very efficient). I suppose the main reason people use strong programs is for analysis, so the improvements should help.

Mark

How do you define 'being better with MultiPV >= 4'? Will just the top line be better, or will the top 4 lines all be better than the corresponding SF MultiPV within the same amount of time?

[lkaufman]

Dragon 2 is now the favorite against Stockfish 13 on typical modern hardware.


Rgds

Is it? They seem to imply that with a pv of one the improvement is small, but the difference between Dragon 1 and SF 13 is 79 elo:

http://www.cegt.net/40_40%20Rating%20Li ... liste.html

The above is a fragment of the whole sentence. Larry wrote "We believe that the MultiPV search is now more effective than the Stockfish MultiPV search, and that if MultiPV is set to more than 4 Dragon 2 is now the favorite against Stockfish 13 on typical modern hardware."

So if you set both Dragon 2 and Stockfish to use MultiPV of 4 or more, Dragon 2 will be stronger in games, which basically means you will get better analysis quicker with Dragon 2 displaying 4 or more lines. Basically we improved MultiPV a lot in Dragon 2 (and frankly the old MultiPV scheme was not very efficient). I suppose the main reason people use strong programs is for analysis, so the improvements should help.

Mark

How do you define 'being better with MultiPV >= 4'? Will just the top line be better, or will the top 4 lines all be better than the corresponding SF MultiPV within the same amount of time?

Well, the only thing we can easily test is which program wins, which only depends on the top line. It's actually not easy to make statements about the quality of the secondary lines, because I don't think that the quality will be constant in any good modern program. But I can say that the clearly lower quality of secondary lines which is a trademark of MCTS search (whether Komodo or Lc0-type program) doesn't apply to normal Komodo Dragon; the secondary lines should still be analyzed to the full depth with high quality. We could probably make the top line stronger by adding some rule that shortens the search depth of obviously bad lines, but we haven't done that.

[lkaufman]

Komodo Dragon 2 is released today at komodochess.com. As usual, it is free for subscribers, priced the same as Dragon (1), which itself will be marked down 20% as will Komoro 14.1 shortly. Anyone who bought Komodo 14 or later version gets 20% discount from these prices.
The improvements over Dragon 1 are primarily in two areas. MultiPV search (in standard mode) has been completely rewritten, and now is
about a hundred elo stronger than in Dragon 1 (or older Komodo versions). It is sufficiently improved that we no longer advocate using MCTS mode
just for the MultiPV benefits it provides. The MCTS mode is also improved in Dragon 2, dramatically so when using Windows, where the elo gain
over the original Dragon is also about a hundred elo. There is a new UCI option called MCTS Optimism, which when set to a positive number
(default is +10) makes Dragon "trust" its own internal move preferences more while exploring the opponent's choices more; a higher setting
might be appropriate for a human opponent. MCTS mode remains the best option for super-strong but human-like play and for preparing openings for
human opponents who will not have access to engines while playing. The standard mode when not using MultiPV has also been improved, but only by
an estimated ten elo. It is nearly the same as the version that won the recent TCEC Swiss, and is a slight upgrade from the version that beat
Stockfish in the fifty game final match of the TCEC chess960 (FRC) championship.
We believe that the MultiPV search is now more effective than the Stockfish MultiPV search, and that if MultiPV is set to more than 4
Dragon 2 is now the favorite against Stockfish 13 on typical modern hardware.
Dragon 2 comes with an AVX2 version (suitable for most PCs since 2013) and a standard one. The AVX2 version is quite a bit faster and
stronger, and we hope that the testing groups will limit testing to AVX2 machines if at all possible. It probably doesn't matter when running
against Stockfish 12 or later, which also gets a lot out of AVX2, but it makes a big difference against programs not using NNUE. There is no
BMI2 version for Dragon 2 as we have not found any meaningful benefit to BMI2 beyond the benefit of AVX2.
Finally, for those who bought Komodo 14 or 14.1 but not Dragon, the strength gains from Komodo 14 in either MCTS or standard mode are nearly
200 elo, only slightly less from Komodo 14.1. All estimated elo gains are at the CCRL blitz time control of 2' + 1" increment, and use Ordo rather
than BayesElo.

I am just curious, is a 2 pawns odds plus Non Castling for Dragon 2 Vs GM Nakamura worse or better than Knight odds at 2'+1" T/C

Castling rights are worth a bit less than a pawn, while a knight in the opening is worth nearly four pawns. Since three is less than four, knight odds should be the larger handicap. Knight odds is really a huge handicap, so if we can ever win a match from an active grandmaster (not a 73 year old like me!) in Rapid (15' + 10") at knight odds, even with Armageddon rule (odds-giver wins on draws), that would be an historic accomplishment.

[Chessqueen]

Komodo Dragon 2 is released today at komodochess.com. As usual, it is free for subscribers, priced the same as Dragon (1), which itself will be marked down 20% as will Komoro 14.1 shortly. Anyone who bought Komodo 14 or later version gets 20% discount from these prices.
The improvements over Dragon 1 are primarily in two areas. MultiPV search (in standard mode) has been completely rewritten, and now is
about a hundred elo stronger than in Dragon 1 (or older Komodo versions). It is sufficiently improved that we no longer advocate using MCTS mode
just for the MultiPV benefits it provides. The MCTS mode is also improved in Dragon 2, dramatically so when using Windows, where the elo gain
over the original Dragon is also about a hundred elo. There is a new UCI option called MCTS Optimism, which when set to a positive number
(default is +10) makes Dragon "trust" its own internal move preferences more while exploring the opponent's choices more; a higher setting
might be appropriate for a human opponent. MCTS mode remains the best option for super-strong but human-like play and for preparing openings for
human opponents who will not have access to engines while playing. The standard mode when not using MultiPV has also been improved, but only by
an estimated ten elo. It is nearly the same as the version that won the recent TCEC Swiss, and is a slight upgrade from the version that beat
Stockfish in the fifty game final match of the TCEC chess960 (FRC) championship.
We believe that the MultiPV search is now more effective than the Stockfish MultiPV search, and that if MultiPV is set to more than 4
Dragon 2 is now the favorite against Stockfish 13 on typical modern hardware.
Dragon 2 comes with an AVX2 version (suitable for most PCs since 2013) and a standard one. The AVX2 version is quite a bit faster and
stronger, and we hope that the testing groups will limit testing to AVX2 machines if at all possible. It probably doesn't matter when running
against Stockfish 12 or later, which also gets a lot out of AVX2, but it makes a big difference against programs not using NNUE. There is no
BMI2 version for Dragon 2 as we have not found any meaningful benefit to BMI2 beyond the benefit of AVX2.
Finally, for those who bought Komodo 14 or 14.1 but not Dragon, the strength gains from Komodo 14 in either MCTS or standard mode are nearly
200 elo, only slightly less from Komodo 14.1. All estimated elo gains are at the CCRL blitz time control of 2' + 1" increment, and use Ordo rather
than BayesElo.

I am just curious, is a 2 pawns odds plus Non Castling for Dragon 2 Vs GM Nakamura worse or better than Knight odds at 2'+1" T/C

Castling rights are worth a bit less than a pawn, while a knight in the opening is worth nearly four pawns. Since three is less than four, knight odds should be the larger handicap. Knight odds is really a huge handicap, so if we can ever win a match from an active grandmaster (not a 73 year old like me!) in Rapid (15' + 10") at knight odds, even with Armageddon rule (odds-giver wins on draws), that would be an historic accomplishment.

That sound very logical but it is hard to believe that a Knight = 3.75 pawn is harder than giving the d, e, and f pawn plus the right to castle

[lkaufman]

Komodo Dragon 2 is released today at komodochess.com. As usual, it is free for subscribers, priced the same as Dragon (1), which itself will be marked down 20% as will Komoro 14.1 shortly. Anyone who bought Komodo 14 or later version gets 20% discount from these prices.
The improvements over Dragon 1 are primarily in two areas. MultiPV search (in standard mode) has been completely rewritten, and now is
about a hundred elo stronger than in Dragon 1 (or older Komodo versions). It is sufficiently improved that we no longer advocate using MCTS mode
just for the MultiPV benefits it provides. The MCTS mode is also improved in Dragon 2, dramatically so when using Windows, where the elo gain
over the original Dragon is also about a hundred elo. There is a new UCI option called MCTS Optimism, which when set to a positive number
(default is +10) makes Dragon "trust" its own internal move preferences more while exploring the opponent's choices more; a higher setting
might be appropriate for a human opponent. MCTS mode remains the best option for super-strong but human-like play and for preparing openings for
human opponents who will not have access to engines while playing. The standard mode when not using MultiPV has also been improved, but only by
an estimated ten elo. It is nearly the same as the version that won the recent TCEC Swiss, and is a slight upgrade from the version that beat
Stockfish in the fifty game final match of the TCEC chess960 (FRC) championship.
We believe that the MultiPV search is now more effective than the Stockfish MultiPV search, and that if MultiPV is set to more than 4
Dragon 2 is now the favorite against Stockfish 13 on typical modern hardware.
Dragon 2 comes with an AVX2 version (suitable for most PCs since 2013) and a standard one. The AVX2 version is quite a bit faster and
stronger, and we hope that the testing groups will limit testing to AVX2 machines if at all possible. It probably doesn't matter when running
against Stockfish 12 or later, which also gets a lot out of AVX2, but it makes a big difference against programs not using NNUE. There is no
BMI2 version for Dragon 2 as we have not found any meaningful benefit to BMI2 beyond the benefit of AVX2.
Finally, for those who bought Komodo 14 or 14.1 but not Dragon, the strength gains from Komodo 14 in either MCTS or standard mode are nearly
200 elo, only slightly less from Komodo 14.1. All estimated elo gains are at the CCRL blitz time control of 2' + 1" increment, and use Ordo rather
than BayesElo.

I am just curious, is a 2 pawns odds plus Non Castling for Dragon 2 Vs GM Nakamura worse or better than Knight odds at 2'+1" T/C

Castling rights are worth a bit less than a pawn, while a knight in the opening is worth nearly four pawns. Since three is less than four, knight odds should be the larger handicap. Knight odds is really a huge handicap, so if we can ever win a match from an active grandmaster (not a 73 year old like me!) in Rapid (15' + 10") at knight odds, even with Armageddon rule (odds-giver wins on draws), that would be an historic accomplishment.

That sound very logical but it is hard to believe that a Knight = 3.75 pawn is harder than giving the d, e, and f pawn plus the right to castle

Dragon 2 give about -4.27 for knight (b1) odds, and about -4.53 for your three pawns and castling handicap, so it agrees with you that knight odds is slightly smaller (about a quarter pawn). The choice of pawns matters, and if you can't castle anyway you would rather give up side pawns than center pawns since the side pawns won't shield a castled king in this case. The good point about knight odds, especially b1, is that it changes the opening and the game much less than removing multiple pawns does. It's much easier for a human to play the computer with knight odds, where normal openings are fine, than with multiple pawns removed, which forces the human to think for himself right from the first move. That's why it's been so difficult for computers to win at knight odds against strong players generally speaking, and why beating a GM in such a match would be a significant accomplishment (in Rapid, not blitz).

[Zerbinati]

@lkaufman i wanted to inform you that from later tests it is emerged that the problem is caused by the GUI version
and not from Dragon 2.

[lkaufman]

@lkaufman i wanted to inform you that from later tests it is emerged that the problem is caused by the GUI version
and not from Dragon 2.

Thanks. Which GUI version was that, in case others have similar problems?

[criko]

Hi Larry,

the M1 version for Mac does not run.
Any hints?

regards
Chris

[lkaufman]

Hi Larry,

the M1 version for Mac does not run.
Any hints?

regards
Chris

Yes, we're working on a fix for the M1 version problem, but if the Mac AVX2 version runs on your machine, that should be nearly as fast; the difference might turn out to be too small to justify separate versions.