Re: [computer-go] Fuego parameter question
If you really want to test MFOG against Fuego, it is better to run Fuego on a strong Linux-machine. The Cygwin-version is significantly slower than the full-build I have on the same machine with Fedora. I provide the Cygwin for people who are not familiar enough with linux or are not able to build the engines themselves with Cygwin. --- With kind regards, Ben Lambrechts Windows builds for GNU Go and Fuego : http://gnugo.baduk.org/ Fuego opening books : http://gnugo.baduk.org/fuegoob.htm On Sat, Dec 5, 2009 at 5:54 AM, David Fotland fotl...@smart-games.comwrote: Many Faces is getting too strong for Gnugo. I test using 8K playouts per move on 19x19 and win about 89% of the games. I just tried testing against Fuego to get a stronger opponent. I used fuego-svn985 from http://gnugo.baduk.org/, already built for Windows. I ran it with: fuego c:\go\goprograms\fuego-svn985\fuego -srand 0 -quiet -config config.txt config.txt is: uct_param_player ignore_clock 1 uct_param_player max_games 8000 uct_param_search number_threads 1 uct_command_player ponder 0 I expected to win 50 to 60% of the games, but won 88% of 1300 games. There were several games that Fuego lost due to a superko violation. Am I missing a parameter to set the rules to Chinese with superko? Am I missing a parameter to give the strength I've seen in KGS tournaments? Perhaps Many Faces is relatively stonger with few playouts due to its knowledge and Fuego will do better with more playouts? David ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Fuego parameter question
On Fri, Dec 04, 2009 at 08:54:39PM -0800, David Fotland wrote: Many Faces is getting too strong for Gnugo. I test using 8K playouts per move on 19x19 and win about 89% of the games. Is there a specific reason for insisting on even games? I still sometimes do tests on 9x9 (e.g. some slight re-tuning for tomorrow tournament) and I'm simply testing with giving gnugo black with no komi; I was actually already thinking about giving reverse komi to get even more accurate results. Petr Pasky Baudis ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Fuego parameter question
Has anyone tried running Fuego on linux via VM on Windows? Is it faster than cygwin-Fuego on Windows? (I don't know how much VM performance varies but http://www.virtualbox.org/ is freely available for testing.) With a newer version of Windows and a multi-core machine it might not a bad option. Ben. From: Ben Lambrechts benedic...@fedoraproject.org To: computer-go computer-go@computer-go.org Sent: Sat, December 5, 2009 3:44:44 AM Subject: Re: [computer-go] Fuego parameter question If you really want to test MFOG against Fuego, it is better to run Fuego on a strong Linux-machine. The Cygwin-version is significantly slower than the full-build I have on the same machine with Fedora. I provide the Cygwin for people who are not familiar enough with linux or are not able to build the engines themselves with Cygwin. --- With kind regards, Ben Lambrechts Windows builds for GNU Go and Fuego : http://gnugo.baduk.org/ Fuego opening books : http://gnugo.baduk.org/fuegoob.htm On Sat, Dec 5, 2009 at 5:54 AM, David Fotland fotl...@smart-games.com wrote: Many Faces is getting too strong for Gnugo. I test using 8K playouts per move on 19x19 and win about 89% of the games. I just tried testing against Fuego to get a stronger opponent. I used fuego-svn985 from http://gnugo.baduk.org/, already built for Windows. I ran it with: fuego c:\go\goprograms\fuego-svn985\fuego -srand 0 -quiet -config config.txt config.txt is: uct_param_player ignore_clock 1 uct_param_player max_games 8000 uct_param_search number_threads 1 uct_command_player ponder 0 I expected to win 50 to 60% of the games, but won 88% of 1300 games. There were several games that Fuego lost due to a superko violation. Am I missing a parameter to set the rules to Chinese with superko? Am I missing a parameter to give the strength I've seen in KGS tournaments? Perhaps Many Faces is relatively stonger with few playouts due to its knowledge and Fuego will do better with more playouts? David ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Fuego parameter question
Ben Lambrechts wrote: If you really want to test MFOG against Fuego, it is better to run Fuego on a strong Linux-machine. The Cygwin-version is significantly slower than the full-build I have on the same machine with Fedora. Isn't that just a matter of using a cygwin version with the same version of GCC? There's no reason why something like a go engine should care what the underlying OS is. -- GCP ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Fuego parameter question
On Sat, Dec 05, 2009 at 04:41:49PM +0100, Gian-Carlo Pascutto wrote: There's no reason why something like a go engine should care what the underlying OS is. Not until it gets multi-threaded, AFAIK pthread emulation by cygwin is far from optimal because threading model in Win32 is very different. Petr Pasky Baudis ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
RE: [computer-go] Fuego parameter question
I'm not testing Fuego against MFGO. I'm using Fuego as part of the MFGO regression tests in place of Gnugo. My test machines are all Windows, as are all of my test scripts. So I don't need the strongest Fuego. I just want a fast, strong program that is better than gnugo. David From: computer-go-boun...@computer-go.org [mailto:computer-go-boun...@computer-go.org] On Behalf Of Ben Lambrechts Sent: Saturday, December 05, 2009 1:45 AM To: computer-go Subject: Re: [computer-go] Fuego parameter question If you really want to test MFOG against Fuego, it is better to run Fuego on a strong Linux-machine. The Cygwin-version is significantly slower than the full-build I have on the same machine with Fedora. I provide the Cygwin for people who are not familiar enough with linux or are not able to build the engines themselves with Cygwin. --- With kind regards, Ben Lambrechts Windows builds for GNU Go and Fuego : http://gnugo.baduk.org/ Fuego opening books : http://gnugo.baduk.org/fuegoob.htm On Sat, Dec 5, 2009 at 5:54 AM, David Fotland fotl...@smart-games.com wrote: Many Faces is getting too strong for Gnugo. I test using 8K playouts per move on 19x19 and win about 89% of the games. I just tried testing against Fuego to get a stronger opponent. I used fuego-svn985 from http://gnugo.baduk.org/, already built for Windows. I ran it with: fuego c:\go\goprograms\fuego-svn985\fuego -srand 0 -quiet -config config.txt config.txt is: uct_param_player ignore_clock 1 uct_param_player max_games 8000 uct_param_search number_threads 1 uct_command_player ponder 0 I expected to win 50 to 60% of the games, but won 88% of 1300 games. There were several games that Fuego lost due to a superko violation. Am I missing a parameter to set the rules to Chinese with superko? Am I missing a parameter to give the strength I've seen in KGS tournaments? Perhaps Many Faces is relatively stonger with few playouts due to its knowledge and Fuego will do better with more playouts? David ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
RE: [computer-go] Fuego parameter question
Good suggestion. I was getting a little worried about overtuning against Gnugo, since it's the only program I've ever tested against. David -Original Message- From: computer-go-boun...@computer-go.org [mailto:computer-go- boun...@computer-go.org] On Behalf Of Petr Baudis Sent: Saturday, December 05, 2009 2:40 AM To: computer-go Subject: Re: [computer-go] Fuego parameter question On Fri, Dec 04, 2009 at 08:54:39PM -0800, David Fotland wrote: Many Faces is getting too strong for Gnugo. I test using 8K playouts per move on 19x19 and win about 89% of the games. Is there a specific reason for insisting on even games? I still sometimes do tests on 9x9 (e.g. some slight re-tuning for tomorrow tournament) and I'm simply testing with giving gnugo black with no komi; I was actually already thinking about giving reverse komi to get even more accurate results. Petr Pasky Baudis ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Fuego parameter question
uct_param_player ignore_clock 1 uct_param_player max_games 8000 uct_param_search number_threads 1 uct_command_player ponder 0 I learnt the other day that ignore_clock only ignores the game time settings, but there is still a default 10 seconds per move. To get rid of that add: go_param timelimit 99 You also need to set max_nodes quite high or Fuego will keep stopping to clear out its tree. I'm setting it to max_games*50, so for 8000: uct_param_search max_nodes 40 According to my notes fuego uses 75M + (65M per million max_nodes). So 15 million nodes will use about 1Gb. (That is on 32-bit linux.) I expected to win 50 to 60% of the games, but won 88% of 1300 games. There were several games that Fuego lost due to a superko violation. Am I missing a parameter to set the rules to Chinese with superko? go_rules chinese (I think the tromp taylor, but I'm surprised superko is not part of it) You can also use go_param_rules to set each rule aspect separately. (BTW, I find using gogui is the best way to understand the fuego settings.) Am I missing a parameter to give the strength I've seen in KGS tournaments? Perhaps Many Faces is relatively stonger with few playouts due to its knowledge and Fuego will do better with more playouts? Can you compare the total CPU time spent on a game by each of Fuego and Many Faces and if Fuego is using less then increase max_games accordingly? Or, given that you just want a strong opponent for regression testing, forget cpu time and simply keep doubling max_games until you reach 50% :-) To answer your question I also have these in my config, which I got from [1]. uct_param_search lock_free 1 uct_param_search virtual_loss 1 The first makes it stronger when using multiple threads. I'm not sure what the second is doing... Darren [1]:http://www.cs.ualberta.ca/TechReports/2009/TR09-09/TR09-09.pdf -- Darren Cook, Software Researcher/Developer http://dcook.org/gobet/ (Shodan Go Bet - who will win?) http://dcook.org/mlsn/ (Multilingual open source semantic network) http://dcook.org/work/ (About me and my work) http://dcook.org/blogs.html (My blogs and articles) ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] KCC won the 3rd UEC Cup
Rank NameAuthor Score* ... 11Katsunari Shinich Sei 2 Katsunari used alpha-beta search and the others used MCTS. Katsunari came second at the UEC cup, but last at GPW Cup. I know there was some grumbling over the seeding at the UEC Cup but that is still quite a difference. Did it have technical problems at Hakone? Darren -- Darren Cook, Software Researcher/Developer http://dcook.org/gobet/ (Shodan Go Bet - who will win?) http://dcook.org/mlsn/ (Multilingual open source semantic network) http://dcook.org/work/ (About me and my work) http://dcook.org/blogs.html (My blogs and articles) ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Fuego parameter question
On Sun, Dec 06, 2009 at 09:05:27AM +0900, Darren Cook wrote: I expected to win 50 to 60% of the games, but won 88% of 1300 games. There were several games that Fuego lost due to a superko violation. Am I missing a parameter to set the rules to Chinese with superko? go_rules chinese (I think the tromp taylor, but I'm surprised superko is not part of it) But it is! :) 6. A turn is either a pass; or a move that doesn't repeat an earlier grid coloring. uct_param_search lock_free 1 uct_param_search virtual_loss 1 The first makes it stronger when using multiple threads. Actually, mainly faster - there should be no strength difference if you limit games number, not time. I'm not sure what the second is doing... You should watch my presentation. ;-) It is a technique for tree search to try to spread parallel-descending threads to different nodes by adding a loss to the node when the thread descends through it and removing it when it is returning back to the root. There is some search efficiency bonus, though the strength gain was very small in my measurements. -- Petr Pasky Baudis A lot of people have my books on their bookshelves. That's the problem, they need to read them. -- Don Knuth ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
RE: [computer-go] Fuego parameter question
Thanks. I tried giving both fuego and MFGO 16K playouts and stopped at with MFGO winning 123/135 = 91% +- 4%. I'm starting again with your suggestions: uct_param_player ignore_clock 1 uct_param_player max_games 16000 uct_param_search number_threads 1 uct_param_player ponder 0 go_param_rules ko_rule pos_superko go_param timelimit 99 uct_param_search max_nodes 40 uct_param_search lock_free 1 uct_param_search virtual_loss 1 David -Original Message- From: computer-go-boun...@computer-go.org [mailto:computer-go- boun...@computer-go.org] On Behalf Of Darren Cook Sent: Saturday, December 05, 2009 4:05 PM To: computer-go Subject: Re: [computer-go] Fuego parameter question uct_param_player ignore_clock 1 uct_param_player max_games 8000 uct_param_search number_threads 1 uct_command_player ponder 0 I learnt the other day that ignore_clock only ignores the game time settings, but there is still a default 10 seconds per move. To get rid of that add: go_param timelimit 99 You also need to set max_nodes quite high or Fuego will keep stopping to clear out its tree. I'm setting it to max_games*50, so for 8000: uct_param_search max_nodes 40 According to my notes fuego uses 75M + (65M per million max_nodes). So 15 million nodes will use about 1Gb. (That is on 32-bit linux.) I expected to win 50 to 60% of the games, but won 88% of 1300 games. There were several games that Fuego lost due to a superko violation. Am I missing a parameter to set the rules to Chinese with superko? go_rules chinese (I think the tromp taylor, but I'm surprised superko is not part of it) You can also use go_param_rules to set each rule aspect separately. (BTW, I find using gogui is the best way to understand the fuego settings.) Am I missing a parameter to give the strength I've seen in KGS tournaments? Perhaps Many Faces is relatively stonger with few playouts due to its knowledge and Fuego will do better with more playouts? Can you compare the total CPU time spent on a game by each of Fuego and Many Faces and if Fuego is using less then increase max_games accordingly? Or, given that you just want a strong opponent for regression testing, forget cpu time and simply keep doubling max_games until you reach 50% :-) To answer your question I also have these in my config, which I got from [1]. uct_param_search lock_free 1 uct_param_search virtual_loss 1 The first makes it stronger when using multiple threads. I'm not sure what the second is doing... Darren [1]:http://www.cs.ualberta.ca/TechReports/2009/TR09-09/TR09-09.pdf -- Darren Cook, Software Researcher/Developer http://dcook.org/gobet/ (Shodan Go Bet - who will win?) http://dcook.org/mlsn/ (Multilingual open source semantic network) http://dcook.org/work/ (About me and my work) http://dcook.org/blogs.html (My blogs and articles) ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] KCC won the 3rd UEC Cup
Darren Cook: 4b1af9bd.3020...@dcook.org: Rank NameAuthor Score* ... 11 Katsunari Shinich Sei 2 Katsunari used alpha-beta search and the others used MCTS. Katsunari came second at the UEC cup, but last at GPW Cup. I know there was some grumbling over the seeding at the UEC Cup but that is still quite a difference. Did it have technical problems at Hakone? Not at all. I guess Katsunari is still competitive on 19 x 19 but not on 9 x 9. Hideki -- g...@nue.ci.i.u-tokyo.ac.jp (Kato) ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/