Crazyhouse Computer Championships 2016

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

Moderators: hgm, Rebel, chrisw

Ferdy
Posts: 4833
Joined: Sun Aug 10, 2008 3:15 pm
Location: Philippines

Crazyhouse Computer Championships 2016

Post by Ferdy »

This will be started this coming 2016-12-02, to the authors you may contact me if you have latest engines to participate.

I will use latest released version of Stockfish variant from below link, or as authors may wish.
https://github.com/niklasf/Stockfish/releases

Tentative pairings based on blitz rating list.
http://ccva.challonge.com/1st_ccva_comp_champ_2016

Conditions and other info.
https://sites.google.com/site/zhassocia ... mpionships
User avatar
MikeB
Posts: 4889
Joined: Thu Mar 09, 2006 6:34 am
Location: Pen Argyl, Pennsylvania

Re: Crazyhouse Computer Championships 2016

Post by MikeB »

Just make sure the rules are clear and unambiguous ...lol
Ferdy
Posts: 4833
Joined: Sun Aug 10, 2008 3:15 pm
Location: Philippines

Re: Crazyhouse Computer Championships 2016

Post by Ferdy »

MikeB wrote:Just make sure the rules are clear and unambiguous ...lol
Tell me what are difficult to understand, I am trying to improve it :).
Ferdy
Posts: 4833
Joined: Sun Aug 10, 2008 3:15 pm
Location: Philippines

Re: Crazyhouse Computer Championships 2016

Post by Ferdy »

Setup and conditions are updated:

Code: Select all

I Stage 1
    1. Group A
        a. Players with odd rank numbers from blitz rating list 
        b. Double elimination format
        c. TC 60m + 10s inc/move
        d. Best of 24 games per match
        e. Tie-break: blitz TC 3m + 2s inc/move, best of 32 games
        f. Top 2 players will advance to stage 2
 
    2. Group B
        a. Players with even rank numbers from blitz rating list 
        b. Double elimination format
        c. TC 60m + 10s inc/move
        d. Best of 24 games per match
        e. Tie-break: blitz TC 3m + 2s inc/move, best of 32 games
        f. Top 2 players will advance to stage 2

II Stage 2
    1. Semi-final
        a. Four players from stage 1
        b. Crossover semi-final will be applied, group A top 1 vs group B top 2
           and group B top 1 vs group A top 2
        c. TC 60m + 10s inc/move
        d. Best of 32 games per match
        e. Tie-break: blitz TC 3m + 2s inc/move best of 32 games
        f. Winners will advance to final i
        g. Losers will advance to final ii
 
    2. Final
        i. The match to determine 1st and 2nd placers
            a. winners stage 2 semi-final
            b. TC 90m + 30s inc/move
            c. Best of 64 games
            d. Tie-break: blitz TC 3m + 2s inc/move best of 64 games
            e. The winner will be the 1st placer or champion
            f. The loser will be the 2nd placer
  
        ii. The match to determine the 3rd and 4th placers
            a. Losers from stage 2 semi-final
            b. TC 60m + 30s inc/move
            c. Best of 32 games
            d. Tie-break: blitz TC 3m + 2s inc/move best of 64 games
            e. The winner will be the 3rd placer
            f. The loser will be the 4th placer
  
III Hardware
    1. Processor: i7-2600K, 3.2 Ghz, 12 GB RAM, 4-physical core, 8-logical core
    2. OS: Windows 7 Professional
    3. Hyper-threading will be set to off

IV Engine settings
    1. Each engine are set to use 1 thread or 1 core
    2. Hash is set to 512 MB
    3. Ponder, own book and learning are set to off

V Starting position
    1. Using ccva-410start.pgn (may still change until 2016-12-01)
    2. Start position in every game in every match will be randomly
       selected by Cute Chess GUI
    3. Each position will be used twice i.e each engine will take starting
       side to move of the position
    
VI Match GUI
    1. Cute Chess
    2. Concurrent game is set to 3
    3. Draw adjudication is set to off
    4. Resign adjudication is set to off
    5. Ponder is set to off
    6. Opening suite depth is set to 100 plies
    7. Opening suite opening order is set to random
 
VII Tables and results
    1. http://ccva.challonge.com/1st_ccva_comp_champ_2016
 
VIII Tournament address
    1. https://sites.google.com/site/zhassociation/computers/tournaments/1st-ccva-computer-championships

IX Others
    1. Engine update is accepted but will be used before a match
    2. The Championships will start on 2016-12-02
https://sites.google.com/site/zhassocia ... conditions
User avatar
MikeB
Posts: 4889
Joined: Thu Mar 09, 2006 6:34 am
Location: Pen Argyl, Pennsylvania

Re: Crazyhouse Computer Championships 2016

Post by MikeB »

Ferdy wrote:Setup and conditions are updated:

Code: Select all

I Stage 1
    1. Group A
        a. Players with odd rank numbers from blitz rating list 
        b. Double elimination format
        c. TC 60m + 10s inc/move
        d. Best of 24 games per match
        e. Tie-break: blitz TC 3m + 2s inc/move, best of 32 games
        f. Top 2 players will advance to stage 2
 
    2. Group B
        a. Players with even rank numbers from blitz rating list 
        b. Double elimination format
        c. TC 60m + 10s inc/move
        d. Best of 24 games per match
        e. Tie-break: blitz TC 3m + 2s inc/move, best of 32 games
        f. Top 2 players will advance to stage 2

II Stage 2
    1. Semi-final
        a. Four players from stage 1
        b. Crossover semi-final will be applied, group A top 1 vs group B top 2
           and group B top 1 vs group A top 2
        c. TC 60m + 10s inc/move
        d. Best of 32 games per match
        e. Tie-break: blitz TC 3m + 2s inc/move best of 32 games
        f. Winners will advance to final i
        g. Losers will advance to final ii
 
    2. Final
        i. The match to determine 1st and 2nd placers
            a. winners stage 2 semi-final
            b. TC 90m + 30s inc/move
            c. Best of 64 games
            d. Tie-break: blitz TC 3m + 2s inc/move best of 64 games
            e. The winner will be the 1st placer or champion
            f. The loser will be the 2nd placer
  
        ii. The match to determine the 3rd and 4th placers
            a. Losers from stage 2 semi-final
            b. TC 60m + 30s inc/move
            c. Best of 32 games
            d. Tie-break: blitz TC 3m + 2s inc/move best of 64 games
            e. The winner will be the 3rd placer
            f. The loser will be the 4th placer
  
III Hardware
    1. Processor: i7-2600K, 3.2 Ghz, 12 GB RAM, 4-physical core, 8-logical core
    2. OS: Windows 7 Professional
    3. Hyper-threading will be set to off

IV Engine settings
    1. Each engine are set to use 1 thread or 1 core
    2. Hash is set to 512 MB
    3. Ponder, own book and learning are set to off

V Starting position
    1. Using ccva-410start.pgn (may still change until 2016-12-01)
    2. Start position in every game in every match will be randomly
       selected by Cute Chess GUI
    3. Each position will be used twice i.e each engine will take starting
       side to move of the position
    
VI Match GUI
    1. Cute Chess
    2. Concurrent game is set to 3
    3. Draw adjudication is set to off
    4. Resign adjudication is set to off
    5. Ponder is set to off
    6. Opening suite depth is set to 100 plies
    7. Opening suite opening order is set to random
 
VII Tables and results
    1. http://ccva.challonge.com/1st_ccva_comp_champ_2016
 
VIII Tournament address
    1. https://sites.google.com/site/zhassociation/computers/tournaments/1st-ccva-computer-championships

IX Others
    1. Engine update is accepted but will be used before a match
    2. The Championships will start on 2016-12-02
https://sites.google.com/site/zhassocia ... conditions
I like it. What does this mean:"Opening suite depth is set to 100 plies" seems like a lot of "plies" - maybe you mean 100 different openings?
Ferdy
Posts: 4833
Joined: Sun Aug 10, 2008 3:15 pm
Location: Philippines

Re: Crazyhouse Computer Championships 2016

Post by Ferdy »

MikeB wrote:I like it. What does this mean:"Opening suite depth is set to 100 plies" seems like a lot of "plies" - maybe you mean 100 different openings?
A Cute Chess setting.

Image
User avatar
MikeB
Posts: 4889
Joined: Thu Mar 09, 2006 6:34 am
Location: Pen Argyl, Pennsylvania

Re: Crazyhouse Computer Championships 2016

Post by MikeB »

Ferdy wrote:
MikeB wrote:I like it. What does this mean:"Opening suite depth is set to 100 plies" seems like a lot of "plies" - maybe you mean 100 different openings?
A Cute Chess setting.

Image
Got it. Ferdy - I have become a Crazy House junkie, this is my latest win at Ladder level 1500:

Code: Select all

[Event "Computer Chess Game"]
[Site "Mac-Pro.local"]
[Date "2016.11.21"]
[Round "-"]
[White "michaelbyrne"]
[Black "Stockfish 8 Variant 64 POPCNT (UCI2WB)"]
[Result "1-0"]
[TimeControl "300+7"]
[Variant "crazyhouse"]
[Annotator "1... +0.53"]

1. e4 d5 {+0.53/5 13} 2. e5 Nd7 {-0.74/4 13} 3. d4 e6 {-1.02/4 13} 4. Nf3
c5 {-4.81/4 13} 5. c3 cxd4 {+0.19/4 14} 6. cxd4 P@e4 {-0.05/2 14} 7. Ng5
Be7 {+3.35/5 14} 8. Nxf7 Bb4+ {-3.83/3 14} 9. Bd2 Bxd2+ {-2.55/4 14} 10.
Nxd2 Kxf7 {+2.31/4 14} 11. P@d6 h5 {-1.70/2 14} 12. Be2 N@f4 {+4.52/2 14}
13. O-O Qg5 {+1.27/3 14} 14. P@g3 B@f3 {+1.55/4 14} 15. B@h4 Nxe2+
{+4.90/4 14} 16. Qxe2 Bxe2 {+1.11/5 14} 17. Bxg5 Q@g4 {+9.64/2 14} 18.
N@d8+ Kg6 {+6.39/5 14} 19. Q@f7+ Kxg5 {+0.61/3 14} 20. h4+ Kh6 {-0.90/4 14}
21. Nxe6 Qg6 {+3.70/2 15} 22. Qxg6+ Kxg6 {-2.77/3 15} 23. Nf4+ Kf7
{-1000.02/3 15} 24. P@g6+ Ke8 {-1000.02/9 14} 25. Q@f7+ Kd8
{-1000.01/127 14} 26. Ne6#
{Black resigns} 1-0
I have found sacrifices and body contact checking to be very useful. Note at move 17 SF thought it was winning , but game was over by move 26.
Ferdy
Posts: 4833
Joined: Sun Aug 10, 2008 3:15 pm
Location: Philippines

Re: Crazyhouse Computer Championships 2016

Post by Ferdy »

MikeB wrote:Got it. Ferdy - I have become a Crazy House junkie, this is my latest win at Ladder level 1500:

Code: Select all

[Event "Computer Chess Game"]
[Site "Mac-Pro.local"]
[Date "2016.11.21"]
[Round "-"]
[White "michaelbyrne"]
[Black "Stockfish 8 Variant 64 POPCNT (UCI2WB)"]
[Result "1-0"]
[TimeControl "300+7"]
[Variant "crazyhouse"]
[Annotator "1... +0.53"]

1. e4 d5 {+0.53/5 13} 2. e5 Nd7 {-0.74/4 13} 3. d4 e6 {-1.02/4 13} 4. Nf3
c5 {-4.81/4 13} 5. c3 cxd4 {+0.19/4 14} 6. cxd4 P@e4 {-0.05/2 14} 7. Ng5
Be7 {+3.35/5 14} 8. Nxf7 Bb4+ {-3.83/3 14} 9. Bd2 Bxd2+ {-2.55/4 14} 10.
Nxd2 Kxf7 {+2.31/4 14} 11. P@d6 h5 {-1.70/2 14} 12. Be2 N@f4 {+4.52/2 14}
13. O-O Qg5 {+1.27/3 14} 14. P@g3 B@f3 {+1.55/4 14} 15. B@h4 Nxe2+
{+4.90/4 14} 16. Qxe2 Bxe2 {+1.11/5 14} 17. Bxg5 Q@g4 {+9.64/2 14} 18.
N@d8+ Kg6 {+6.39/5 14} 19. Q@f7+ Kxg5 {+0.61/3 14} 20. h4+ Kh6 {-0.90/4 14}
21. Nxe6 Qg6 {+3.70/2 15} 22. Qxg6+ Kxg6 {-2.77/3 15} 23. Nf4+ Kf7
{-1000.02/3 15} 24. P@g6+ Ke8 {-1000.02/9 14} 25. Q@f7+ Kd8
{-1000.01/127 14} 26. Ne6#
{Black resigns} 1-0
I have found sacrifices and body contact checking to be very useful. Note at move 17 SF thought it was winning , but game was over by move 26.
Contact check is indeed one of the eval features that helps in the evaluation. It can be safe contact check or an unsafe contact check. The latter disrupts the current defense structure of the king.

BTW where do you base the number 1500 in Ladder Level 1500?
User avatar
MikeB
Posts: 4889
Joined: Thu Mar 09, 2006 6:34 am
Location: Pen Argyl, Pennsylvania

Re: Crazyhouse Computer Championships 2016

Post by MikeB »

Ferdy wrote:
MikeB wrote:Got it. Ferdy - I have become a Crazy House junkie, this is my latest win at Ladder level 1500:

Code: Select all

[Event "Computer Chess Game"]
[Site "Mac-Pro.local"]
[Date "2016.11.21"]
[Round "-"]
[White "michaelbyrne"]
[Black "Stockfish 8 Variant 64 POPCNT (UCI2WB)"]
[Result "1-0"]
[TimeControl "300+7"]
[Variant "crazyhouse"]
[Annotator "1... +0.53"]

1. e4 d5 {+0.53/5 13} 2. e5 Nd7 {-0.74/4 13} 3. d4 e6 {-1.02/4 13} 4. Nf3
c5 {-4.81/4 13} 5. c3 cxd4 {+0.19/4 14} 6. cxd4 P@e4 {-0.05/2 14} 7. Ng5
Be7 {+3.35/5 14} 8. Nxf7 Bb4+ {-3.83/3 14} 9. Bd2 Bxd2+ {-2.55/4 14} 10.
Nxd2 Kxf7 {+2.31/4 14} 11. P@d6 h5 {-1.70/2 14} 12. Be2 N@f4 {+4.52/2 14}
13. O-O Qg5 {+1.27/3 14} 14. P@g3 B@f3 {+1.55/4 14} 15. B@h4 Nxe2+
{+4.90/4 14} 16. Qxe2 Bxe2 {+1.11/5 14} 17. Bxg5 Q@g4 {+9.64/2 14} 18.
N@d8+ Kg6 {+6.39/5 14} 19. Q@f7+ Kxg5 {+0.61/3 14} 20. h4+ Kh6 {-0.90/4 14}
21. Nxe6 Qg6 {+3.70/2 15} 22. Qxg6+ Kxg6 {-2.77/3 15} 23. Nf4+ Kf7
{-1000.02/3 15} 24. P@g6+ Ke8 {-1000.02/9 14} 25. Q@f7+ Kd8
{-1000.01/127 14} 26. Ne6#
{Black resigns} 1-0
I have found sacrifices and body contact checking to be very useful. Note at move 17 SF thought it was winning , but game was over by move 26.
Contact check is indeed one of the eval features that helps in the evaluation. It can be safe contact check or an unsafe contact check. The latter disrupts the current defense structure of the king.

BTW where do you base the number 1500 in Ladder Level 1500?
Not a 100% sure I am answering question. First I just copied the ladder option from SF sse. In the sse version , the Ladder rating was initiated at 1200 based on 64 nodes analyzed. Not calculated sceintificaly by any means , just a gut feel on how it played and more importantly how levels 1600 and 1700 played. Base on the premise that doubling of speed , or in this case nodes raises ELO 100 points, which would mean every 25 ELO would increase the nodes roughly by 19% ( the precise number is used the code ) so 1.19^4 ~= 2. So level 1200 is 64 nodes , 1225 is 64 * 1.19, level 1250 is 64 * 1.19^2, level 1300 is 128 nodes ( double level 1200 or 1.19^4 * 64 , all the way up to level 2600 which happens to be about 1 million nodes . Level 1600 is about 1000 nodes which is about the strength of a average club player. Hopefully I answered your question
Ferdy
Posts: 4833
Joined: Sun Aug 10, 2008 3:15 pm
Location: Philippines

Re: Crazyhouse Computer Championships 2016

Post by Ferdy »

MikeB wrote:
Ferdy wrote:
MikeB wrote:Got it. Ferdy - I have become a Crazy House junkie, this is my latest win at Ladder level 1500:

Code: Select all

[Event "Computer Chess Game"]
[Site "Mac-Pro.local"]
[Date "2016.11.21"]
[Round "-"]
[White "michaelbyrne"]
[Black "Stockfish 8 Variant 64 POPCNT (UCI2WB)"]
[Result "1-0"]
[TimeControl "300+7"]
[Variant "crazyhouse"]
[Annotator "1... +0.53"]

1. e4 d5 {+0.53/5 13} 2. e5 Nd7 {-0.74/4 13} 3. d4 e6 {-1.02/4 13} 4. Nf3
c5 {-4.81/4 13} 5. c3 cxd4 {+0.19/4 14} 6. cxd4 P@e4 {-0.05/2 14} 7. Ng5
Be7 {+3.35/5 14} 8. Nxf7 Bb4+ {-3.83/3 14} 9. Bd2 Bxd2+ {-2.55/4 14} 10.
Nxd2 Kxf7 {+2.31/4 14} 11. P@d6 h5 {-1.70/2 14} 12. Be2 N@f4 {+4.52/2 14}
13. O-O Qg5 {+1.27/3 14} 14. P@g3 B@f3 {+1.55/4 14} 15. B@h4 Nxe2+
{+4.90/4 14} 16. Qxe2 Bxe2 {+1.11/5 14} 17. Bxg5 Q@g4 {+9.64/2 14} 18.
N@d8+ Kg6 {+6.39/5 14} 19. Q@f7+ Kxg5 {+0.61/3 14} 20. h4+ Kh6 {-0.90/4 14}
21. Nxe6 Qg6 {+3.70/2 15} 22. Qxg6+ Kxg6 {-2.77/3 15} 23. Nf4+ Kf7
{-1000.02/3 15} 24. P@g6+ Ke8 {-1000.02/9 14} 25. Q@f7+ Kd8
{-1000.01/127 14} 26. Ne6#
{Black resigns} 1-0
I have found sacrifices and body contact checking to be very useful. Note at move 17 SF thought it was winning , but game was over by move 26.
Contact check is indeed one of the eval features that helps in the evaluation. It can be safe contact check or an unsafe contact check. The latter disrupts the current defense structure of the king.

BTW where do you base the number 1500 in Ladder Level 1500?
Not a 100% sure I am answering question. First I just copied the ladder option from SF sse. In the sse version , the Ladder rating was initiated at 1200 based on 64 nodes analyzed. Not calculated sceintificaly by any means , just a gut feel on how it played and more importantly how levels 1600 and 1700 played. Base on the premise that doubling of speed , or in this case nodes raises ELO 100 points, which would mean every 25 ELO would increase the nodes roughly by 19% ( the precise number is used the code ) so 1.19^4 ~= 2. So level 1200 is 64 nodes , 1225 is 64 * 1.19, level 1250 is 64 * 1.19^2, level 1300 is 128 nodes ( double level 1200 or 1.19^4 * 64 , all the way up to level 2600 which happens to be about 1 million nodes . Level 1600 is about 1000 nodes which is about the strength of a average club player. Hopefully I answered your question
Is SF sse a zh variant too?

I plan to run Sf zh with ladder to different engines I have in blitz rating list, which is based on Sjeng 11.2 set at rating 2300 - my estimate of its blitz fics rating.