SF 111113 vs. Houdini 3 match, TC 30 min + 30 sec

Discussion of computer chess matches and engine tournaments.

Moderators: hgm, Rebel, chrisw

ouachita
Posts: 454
Joined: Tue Jan 15, 2013 4:33 pm
Location: Ritz-Carlton, NYC
Full name: Bobby Johnson

Re: SF 111113 vs. Houdini 4 match, TC 30 min + 30 sec

Post by ouachita »

got it; it's not relevant to IA but is relevant to engine v engine competition on my computer. But what about when using Deep Fritz "Deep Position Analysis" which analyses both sides of a game for a predetermined number of moves or time or depth? It likely is used then, right?
SIM, PhD, MBA, PE
majortom
Posts: 669
Joined: Mon Nov 04, 2013 10:19 pm

Re: SF 111113 vs. Houdini 4 match, TC 30 min + 30 sec

Post by majortom »

ouachita wrote:got it; it's not relevant to IA but is relevant to engine v engine competition on my computer. But what about when using Deep Fritz "Deep Position Analysis" which analyses both sides of a game for a predetermined number of moves or time or depth? It likely is used then, right?
Have no idea, I never used "Deep Position Analysis", sorry.
majortom
Posts: 669
Joined: Mon Nov 04, 2013 10:19 pm

Re: SF 111113 vs. Houdini 4 match, TC 30 min + 30 sec

Post by majortom »

TC 30'+30"

i7-3960x
1-core
ponder off
colours reversed openings
1024 mb hash
average games length ~2 h 35 m

167 games with Houdini 4:

Code: Select all

 Program                          Elo    +   -   Games   Score   Av.Op.  Draws

1 Stockfish 111113 64 SSE4.2     : 2968   21  21   469    55.1 %   2932   53.5 %
2 Houdini 4 Pro x64              : 2951   36  36   167    47.6 %   2968   53.3 %
3 Houdini 3 Pro x64              : 2922   27  27   302    43.4 %   2968   53.6 %

Code: Select all

1 Stockfish 111113 64 SSE4.2     : 2968  469 (+133,=251,- 85), 55.1 %
2 Houdini 4 Pro x64              : 2951  167 (+ 35,= 89,- 43), 47.6 %
3 Houdini 3 Pro x64              : 2922  302 (+ 50,=162,- 90), 43.4 %
Games

To be continued...
majortom
Posts: 669
Joined: Mon Nov 04, 2013 10:19 pm

Re: SF 111113 vs. Houdini 4 match, TC 30 min + 30 sec

Post by majortom »

TC 30'+30"

i7-3960x
1-core
ponder off
colours reversed openings
1024 mb hash
average games length ~2 h 35 m

190 games with Houdini 4:

Code: Select all

 Program                          Elo    +   -   Games   Score   Av.Op.  Draws

1 Stockfish 111113 64 SSE4.2     : 2416   21  21   492    54.7 %   2384   54.1 %
2 Houdini 4 Pro x64              : 2405   33  33   190    48.4 %   2416   54.7 %
3 Houdini 3 Pro x64              : 2370   27  27   302    43.4 %   2416   53.6 %

Code: Select all

1 Stockfish 111113 64 SSE4.2     : 2416  492 (+136,=266,- 90), 54.7 %
2 Houdini 4 Pro x64              : 2405  190 (+ 40,=104,- 46), 48.4 %
3 Houdini 3 Pro x64              : 2370  302 (+ 50,=162,- 90), 43.4 %
Games

To be continued...
beram
Posts: 1187
Joined: Wed Jan 06, 2010 3:11 pm

Re: SF 111113 vs. Houdini 4 match, TC 30 min + 30 sec

Post by beram »

Thx for testing at LTC on fast hardware
Question: why do you use such a small openingbook?
Almost every game has kind of stonewall positions with fianchetto bishops on g2 or both on g2 and g7. Its all 1 g3 or English kind of openings A30, A04, B20 and so on
More variety would make far more interesting games and perhaps also other results
majortom
Posts: 669
Joined: Mon Nov 04, 2013 10:19 pm

Re: SF 111113 vs. Houdini 4 match, TC 30 min + 30 sec

Post by majortom »

beram wrote:Thx for testing at LTC on fast hardware
Question: why do you use such a small openingbook?
Almost every game has kind of stonewall positions with fianchetto bishops on g2 or both on g2 and g7. Its all 1 g3 or English kind of openings A30, A04, B20 and so on
More variety would make far more interesting games and perhaps also other results
Thank you for your interest, Bram!
Do you mean a small book or similar positions?
There are 1406 positions, but only A&B ECO.
I haven't checked this book before the start.
When 302 games will be played I will replace that book.
What book do you prefer?
Need the book with 349 positions at least.
beram
Posts: 1187
Joined: Wed Jan 06, 2010 3:11 pm

Re: SF 111113 vs. Houdini 4 match, TC 30 min + 30 sec

Post by beram »

majortom wrote:
beram wrote:Thx for testing at LTC on fast hardware
Question: why do you use such a small openingbook?
Almost every game has kind of stonewall positions with fianchetto bishops on g2 or both on g2 and g7. Its all 1 g3 or English kind of openings A30, A04, B20 and so on
More variety would make far more interesting games and perhaps also other results
Thank you for your interest, Bram!
Do you mean a small book or similar positions?
There are 1406 positions, but only A&B ECO.
I haven't checked this book before the start.
When 302 games will be played I will replace that book.
What book do you prefer?
Need the book with 349 positions at least.
No specific book in particular but a wide spread variety
For example I like Silver testsuite or Noomen 2012 testuite

thx for answering
majortom
Posts: 669
Joined: Mon Nov 04, 2013 10:19 pm

Re: SF 111113 vs. Houdini 4 match, TC 30 min + 30 sec

Post by majortom »

beram wrote:
majortom wrote:
beram wrote:Thx for testing at LTC on fast hardware
Question: why do you use such a small openingbook?
Almost every game has kind of stonewall positions with fianchetto bishops on g2 or both on g2 and g7. Its all 1 g3 or English kind of openings A30, A04, B20 and so on
More variety would make far more interesting games and perhaps also other results
Thank you for your interest, Bram!
Do you mean a small book or similar positions?
There are 1406 positions, but only A&B ECO.
I haven't checked this book before the start.
When 302 games will be played I will replace that book.
What book do you prefer?
Need the book with 349 positions at least.
No specific book in particular but a wide spread variety
For example I like Silver testsuite or Noomen 2012 testuite

thx for answering
Silver gambit suite +
Silver suite +
Salvo Spitaleri opening suite 01.04.2010 +
Sedat Canbaz perfect 2011 top 100 +
Noomen testsuite 2008 + other Noomen's books

- Double games and some similar positions was deleted.

=500 positions

http://treu.ru/pgn/top500_2013.pgn
majortom
Posts: 669
Joined: Mon Nov 04, 2013 10:19 pm

Re: SF 111113 vs. Houdini 4 match, TC 30 min + 30 sec

Post by majortom »

Tournament was stopped after 198 games and started with wide range book (349 positions for 698 games). 52 positions (for 104 games) from previous book was added to the end of new book.
There are 401 positions for 802 games in new book.

http://treu.ru/pgn/SF_H4_2013_401.pgn

Enjoy!
majortom
Posts: 669
Joined: Mon Nov 04, 2013 10:19 pm

Re: SF 111113 vs. Houdini 4 match, TC 30 min + 30 sec

Post by majortom »

TC 30'+30"

i7-3960x
1-core
ponder off
colours reversed openings
1024 mb hash
average games length ~2 h 35 m

220 games with Houdini 4:

Code: Select all

 Program                          Elo    +   -   Games   Score   Av.Op.  Draws

1 Stockfish 111113 64 SSE4.2     : 2964   20  20   522    53.9 %   2936   53.8 %
2 Houdini 4 Pro x64              : 2962   31  31   220    49.8 %   2964   54.1 %
3 Houdini 3 Pro x64              : 2917   27  27   302    43.4 %   2964   53.6 %

Code: Select all

1 Stockfish 111113 64 SSE4.2     : 2964  522 (+141,=281,-100), 53.9 %
2 Houdini 4 Pro x64              : 2962  220 (+ 50,=119,- 51), 49.8 %
3 Houdini 3 Pro x64              : 2917  302 (+ 50,=162,- 90), 43.4 %
Games

To be continued...