SFNNUE Fails!! In this Positions!!

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

Moderators: Harvey Williamson, Dann Corbit, hgm

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
User avatar
Master Om
Posts: 412
Joined: Wed Nov 24, 2010 9:57 am
Location: INDIA

Re: SFNNUE Fails!! In this Positions!!

Post by Master Om » Wed Aug 26, 2020 4:36 pm

peter wrote:
Wed Aug 26, 2020 3:42 pm
Master Om wrote:
Wed Aug 26, 2020 1:24 pm
Also can u provide the links u mentioned above ?
https://github.com/joergoster/Stockfish ... nue_always

https://github.com/syzygy1/Cfish/

Of Cfish you'll have to build your own compile of the latest version, releases are from July only.

https://github.com/amchess/ShashChess/releases/tag/13.1

https://github.com/dkappe/NetRunner/releases/tag/Nr0.1
Thanks...
Always Expect the Unexpected

zullil
Posts: 6442
Joined: Mon Jan 08, 2007 11:31 pm
Location: PA USA
Full name: Louis Zulli

Re: SFNNUE Fails!! In this Positions!!

Post by zullil » Wed Aug 26, 2020 4:40 pm

Master Om wrote:
Wed Aug 26, 2020 4:36 pm
zullil wrote:
Wed Aug 26, 2020 4:23 pm
Master Om wrote:
Wed Aug 26, 2020 2:53 pm

Qxf6 is clearly better move.
Can't imagine the engine would ever lose or draw after Bxa7.

Though Qxf6 might save some energy, I suppose. :D
Thats why we called it "best" not good or better :D
But in your initial post you used "fails". Not sure that finding a move that always wins is a failure. :wink:

peter
Posts: 2171
Joined: Sat Feb 16, 2008 6:38 am
Full name: Peter Martan

Re: SFNNUE Fails!! In this Positions!!

Post by peter » Wed Aug 26, 2020 5:25 pm

zullil wrote:
Wed Aug 26, 2020 4:40 pm
Master Om wrote:
Wed Aug 26, 2020 4:36 pm
zullil wrote:
Wed Aug 26, 2020 4:23 pm
Master Om wrote:
Wed Aug 26, 2020 2:53 pm

Qxf6 is clearly better move.
Can't imagine the engine would ever lose or draw after Bxa7.

Though Qxf6 might save some energy, I suppose. :D
Thats why we called it "best" not good or better :D
But in your initial post you used "fails". Not sure that finding a move that always wins is a failure. :wink:
How about bxa4 then?
:)

Pure NNUE-Eval in depth 42/66, SyzygyProbeLimit 0, SSE4.1-popcnt- compile, 12x3GHz CPU, 32G hash.

704: Corbit,D - Tactical,Insanity
1B1r4/rp2npkp/2b1pbp1/1qp5/nPN1R3/1P1P1QP1/2P2PBP/5R1K w - - 0 1

Analysis by Cfish 250820 x64 POP N:

1. +- (17.56): 1.Dxf6+ Kxf6 2.Le5+ Kg5 3.Lg7 Lxe4 4.f4+ Kh5 5.Lxe4 g5 6.Se5 Dc6 7.g4+ Kh4 8.Lf6 h6 9.fxg5 Sg8 10.Lxd8 Dd6 11.gxh6+ Dxd8 12.Tg1 Kg5 13.Sxf7+ Kf4 14.h7 Sf6 15.Sxd8 Sxh7 16.Lxh7 b6 17.Sxe6+ Ke3 18.Lf5 Sc3 19.bxc5 bxc5 20.g5 Kd2 21.g6 Se2 22.Tf1 Ta8 23.Sxc5 Tf8 24.Se4+ Kxc2 25.Sg3 Sd4 26.Le4 Txf1+ 27.Sxf1 Se6 28.b4 Kd1 29.b5 Ke1 30.Sg3 Sg5 31.b6 Sxe4 32.dxe4 Kd2 33.b7 Kd3

2. +- (6.59): 1.bxa4 Dxb4 2.Lxa7 Lxe4 3.Dxe4 Sc6 4.c3 Lxc3 5.Lb6 Td7 6.De2 Dxa4 7.Lxc5 Td8 8.Lb6 Td7 9.Tb1 Kg8 10.Dd1 Da8 11.Le4 Lg7 12.Le3 Td8 13.Sb6 Da5 14.Df1 Da6 15.h4 Sd4 16.Dg2 Sf5 17.Lxf5 exf5

3. +- (3.45): 1.Lxa7 Lxe4 2.dxe4 e5 3.bxc5 Sxc5 4.De3 Se6 5.Lb6 Ta8 6.Dd2 Ta6 7.Le3 Kg8 8.Lh3 Dc6 9.Lxe6 Dxe6 10.Td1 Sc6 11.Kg2 Ta2 12.Dd7 b5 13.Dxe6 fxe6 14.Sd6 Txc2 15.Sxb5 Tb2 16.Sc7 Kf7 17.Td6 Sd4 18.Td7+ Le7 19.Lg5 Sc6 20.Lxe7 Sxe7 21.Td6 Sg8 22.Sxe6 Ke7 23.Tb6 Sf6 24.Sc5 Sg4 25.Sd3

Net of 24.08. from here

https://tests.stockfishchess.org/nns

Sorry for german abbr., forgot to change.
Peter.

User avatar
Master Om
Posts: 412
Joined: Wed Nov 24, 2010 9:57 am
Location: INDIA

Re: SFNNUE Fails!! In this Positions!!

Post by Master Om » Wed Aug 26, 2020 6:39 pm

zullil wrote:
Wed Aug 26, 2020 4:40 pm
Master Om wrote:
Wed Aug 26, 2020 4:36 pm
zullil wrote:
Wed Aug 26, 2020 4:23 pm
Master Om wrote:
Wed Aug 26, 2020 2:53 pm

Qxf6 is clearly better move.
Can't imagine the engine would ever lose or draw after Bxa7.

Though Qxf6 might save some energy, I suppose. :D
Thats why we called it "best" not good or better :D
But in your initial post you used "fails". Not sure that finding a move that always wins is a failure. :wink:
It fails to find the best move. I spent money in buying hardware and investing time in sw only to make it find the best move in shortest time :-P :D
Always Expect the Unexpected

User avatar
MikeB
Posts: 4500
Joined: Thu Mar 09, 2006 5:34 am
Location: Pen Argyl, Pennsylvania

Re: SFNNUE Fails!! In this Positions!!

Post by MikeB » Thu Aug 27, 2020 1:44 pm

drewdrew wrote:
Wed Aug 26, 2020 3:58 am
MikeB wrote:
Wed Aug 26, 2020 1:35 am

[D]6R1/8/2pB3k/2P4p/5p1q/5P2/4P1K1/8 w - - 0 1

there's probably a shorter mate
info depth 56 seldepth 72 multipv 1 score mate 40 nodes 293405455094 nps 167524134 hashfull 996 tbhits 0 time 1751422 pv d6f8 h6h7 g8g7 h7h8 f8e7 h4e1 e7f6 e1e2 g2h1 e2e6 f6d4 e6c4 d4a1 h5h4 a1f6 c4e6 f6d4 e6c4 d4a1 h4h3 a1f6 c4e6 f6d4 e6c4 d4a1 h3h2 a1f6 c4e6 f6d4 e6h6 g7g6 h8h7 g6h6 h7h6 d4e5 h6g5 e5d6 g5f5 h1h2 f5g5 h2h3 g5f5 h3h4 f5f6 h4g4 f6e6 d6f4 e6d5 f4d6 d5d4 f3f4 d4e4 f4f5 e4d5 g4f3 d5c4 f5f6 c4c3 f6f7 c3d3 f7f8q d3c3 d6e7 c3d3 f8g8 d3c3 e7f6 c3b4 g8b8 b4c5
mate in 37 at most:
info depth 120 seldepth 74 multipv 1 score mate 37 nodes 2294117329989 nps 495657885 hashfull 985 tbhits 129851233105 time 4628429 pv d6f8 h6h7 g8g7 h7h8 f8e7 h4e1 e7f6 e1e2 g2h1 e2e6 f6d4 e6c4 d4a1 h5h4 a1f6 c4e6 f6d4 e6c4 d4a1 h4h3 a1f6 c4e6 f6d4 e6c4 d4a1 h3h2 a1f6 c4e6 f6d4 e6h6 g7g6 h8h7 g6h6 h7h6 d4e5 h6g5 e5d6 g5f5 h1h2 f5f6 d6f4 f6g6 h2h3 g6h5 f4e3 h5g6 h3g4 g6f6 f3f4 f6g6 f4f5 g6f6 g4f4 f6f7 f4e5 f7e7 f5f6 e7f7 e5f5 f7e8 f5e6 e8d8 f6f7 d8c7 f7f8q c7b7 e6d6 b7a6 f8b8 a6a5 e3d2 a5a4 b8b4
that's a big number - please do share what machine /OS - general System information/ you are using...
Image

User avatar
Ovyron
Posts: 4405
Joined: Tue Jul 03, 2007 2:30 am

Re: SFNNUE Fails!! In this Positions!!

Post by Ovyron » Sun Sep 13, 2020 6:48 pm

Master Om wrote:
Wed Aug 26, 2020 6:39 pm
zullil wrote:
Wed Aug 26, 2020 4:40 pm
Master Om wrote:
Wed Aug 26, 2020 4:36 pm
zullil wrote:
Wed Aug 26, 2020 4:23 pm
Master Om wrote:
Wed Aug 26, 2020 2:53 pm

Qxf6 is clearly better move.
Can't imagine the engine would ever lose or draw after Bxa7.

Though Qxf6 might save some energy, I suppose. :D
Thats why we called it "best" not good or better :D
But in your initial post you used "fails". Not sure that finding a move that always wins is a failure. :wink:
It fails to find the best move. I spent money in buying hardware and investing time in sw only to make it find the best move in shortest time :-P :D
Depends on your definition of "best." Mating in fewest moves isn't better than mating with more moves, that's why "find shortest mate" quizes are dumb.

I showcased this in my game against mmt:



So, what's the best move here?

Ne7!!



0-1

So, why was it the best? Because it caused my opponent to resign!

Would he have resigned if I played the move that was mating the fastest? Maybe not, as perhaps he'd have continued the game to see if I continued playing the moves that mated the fastest, or if I deviated later.

But Ne7 was the worst scoring move I could find in the shortest time that still kept the win. Ne7 is a clown move, with more moves like Ne7 coming in the future, my opponent realized it had became a circus and resigned on the spot. That's why Ne7 is best.

So I don't get why Qxf6 is best or why people consider moves that mate the fastest better than others. If you enjoy playing from winning positions on your games then it's clear Bxa7 will produce more of such positions than Qxf6, making it better.

And if Bxa7 is considered like a clown move by your opponent prompting them to resign on the next move, once again, you've won much earlier than playing Qxf6's mating sequence.

Because the game has already been decided at this point, how you win it is a matter of style, and there's no style better than another. Anybody finding playing suboptimal moves that still win on positions where moves that mate sooner are available annoying should realize this is the fault of the losing player that doesn't resign.

User avatar
Master Om
Posts: 412
Joined: Wed Nov 24, 2010 9:57 am
Location: INDIA

Re: SFNNUE Fails!! In this Positions!!

Post by Master Om » Fri Oct 09, 2020 6:10 am

Ovyron wrote:
Sun Sep 13, 2020 6:48 pm
Master Om wrote:
Wed Aug 26, 2020 6:39 pm
zullil wrote:
Wed Aug 26, 2020 4:40 pm
Master Om wrote:
Wed Aug 26, 2020 4:36 pm
zullil wrote:
Wed Aug 26, 2020 4:23 pm
Master Om wrote:
Wed Aug 26, 2020 2:53 pm

Qxf6 is clearly better move.
Can't imagine the engine would ever lose or draw after Bxa7.

Though Qxf6 might save some energy, I suppose. :D
Thats why we called it "best" not good or better :D
But in your initial post you used "fails". Not sure that finding a move that always wins is a failure. :wink:
It fails to find the best move. I spent money in buying hardware and investing time in sw only to make it find the best move in shortest time :-P :D
Depends on your definition of "best." Mating in fewest moves isn't better than mating with more moves, that's why "find shortest mate" quizes are dumb.

I showcased this in my game against mmt:



So, what's the best move here?

Ne7!!



0-1

So, why was it the best? Because it caused my opponent to resign!

Would he have resigned if I played the move that was mating the fastest? Maybe not, as perhaps he'd have continued the game to see if I continued playing the moves that mated the fastest, or if I deviated later.

But Ne7 was the worst scoring move I could find in the shortest time that still kept the win. Ne7 is a clown move, with more moves like Ne7 coming in the future, my opponent realized it had became a circus and resigned on the spot. That's why Ne7 is best.

So I don't get why Qxf6 is best or why people consider moves that mate the fastest better than others. If you enjoy playing from winning positions on your games then it's clear Bxa7 will produce more of such positions than Qxf6, making it better.

And if Bxa7 is considered like a clown move by your opponent prompting them to resign on the next move, once again, you've won much earlier than playing Qxf6's mating sequence.

Because the game has already been decided at this point, how you win it is a matter of style, and there's no style better than another. Anybody finding playing suboptimal moves that still win on positions where moves that mate sooner are available annoying should realize this is the fault of the losing player that doesn't resign.
Depends on your definition of "best." Mating in fewest moves isn't better than mating with more moves, that's why "find shortest mate" quizes are dumb.
Best is that which gives shortest path to win. Mate in few moves is not dumb. Dumb is that who can mate in 2 but mates in 30 thinking a win is a win lol.
So I don't get why Qxf6 is best or why people consider moves that mate the fastest better than others.
U wont get it means its your lack of understanding. Qxf6 has twice the eval than next moves. Its diff with high margin.
Time is always a factor in game play. Be it CC or any other form of chess. Win is always considered either at minimum time or minimum moves. No one trolls to win.
Always Expect the Unexpected

User avatar
Ovyron
Posts: 4405
Joined: Tue Jul 03, 2007 2:30 am

Re: SFNNUE Fails!! In this Positions!!

Post by Ovyron » Mon Oct 12, 2020 5:57 am

Master Om wrote:
Fri Oct 09, 2020 6:10 am
Time is always a factor in game play.
It's only a factor when the result of the game isn't known. When it's known one side has won the game, then time doesn't matter, and it's the responsibility of the losing player to resign when this is the case to save time.

The winning player has already won on the board, and should have the right to do anything they want. They've won even if the game result hasn't appeared yet, it's a matter of... time.

User avatar
Master Om
Posts: 412
Joined: Wed Nov 24, 2010 9:57 am
Location: INDIA

Re: SFNNUE Fails!! In this Positions!!

Post by Master Om » Wed Oct 21, 2020 4:48 pm

Ovyron wrote:
Mon Oct 12, 2020 5:57 am
Master Om wrote:
Fri Oct 09, 2020 6:10 am
Time is always a factor in game play.
It's only a factor when the result of the game isn't known. When it's known one side has won the game, then time doesn't matter, and it's the responsibility of the losing player to resign when this is the case to save time.

The winning player has already won on the board, and should have the right to do anything they want. They've won even if the game result hasn't appeared yet, it's a matter of... time.
It's only a factor when the result of the game isn't known.
Even if known or not known only an idiot chooses the longest path. It doesnot make sense if i can mate in 10 will troll and play for 49 moves to mate as both wins.
then time doesn't matter
Time always matters.
The winning player has already won on the board, and should have the right to do anything they want.
Absolutely not unless he is an idiot. As idiots and do anything :D
Only way is shortest way to win.
it's a matter of... time.
And its always chosen shortest. (Fermat's principle.)
Always Expect the Unexpected

User avatar
Ovyron
Posts: 4405
Joined: Tue Jul 03, 2007 2:30 am

Re: SFNNUE Fails!! In this Positions!!

Post by Ovyron » Sat Oct 24, 2020 10:50 pm

Master Om wrote:
Wed Oct 21, 2020 4:48 pm
Even if known or not known only an idiot chooses the longest path.
Only an idiot would not resign in that situation. If I have mate in 10 against you, and you're not resigning, I can have my fun and mate you in 49 or any amount of moves I desire. I've won the game, it's my choice how I win it.

If you don't like it, just resign and stop wasting everyone's time.

Post Reply