bayeselo games limit ?

[Daniel Shawul]

Is there a limit on the max number of games?. I am doing very fast games and the games pgn builds up quite quickly and it seems bayeselo can't handle more than 10000 games. Is that so ? Also is there a way to make it work by giving it only the number of wins/losses/draws ?

Thanks

[FrancoisK]

I currently use it for millions of games, but they are divided into several pgns. (up to 512000 games per PGN if i am not mistaken).
Never found any limit so far, but it is slow of course as it has to parse PGNs (between half an hour and one hour to read all pgns). I would also be very interested if it could take result grids as input instead of full pgns.

[Daniel Shawul]

500k games is good enough for me but maximum that it reads is 10k even though there are about 120k in a 60MBs file.

Code: Select all

version 0056, Copyright (C) 1997-2007 Remi Coulom.
compiled Jan 30 2007 20:30:07.
This program comes with ABSOLUTELY NO WARRANTY.
This is free software, and you are welcome to redistribute it
under the terms and conditions of the GNU General Public License.
See http://www.gnu.org/copyleft/gpl.html for details.
ResultSet>readpgn test.pgn
10726 game(s) loaded, 0 game(s) with unknown result ignored.
ResultSet>elo
ResultSet-EloRating>mm
00:00:00,00
ResultSet-EloRating>exactdist
00:00:00,18
ResultSet-EloRating>ratings
Rank Name   Elo    +    - games score oppo. draws
   1 AW10    14   16   16  1055   53%    -4   33%
   2 AW14     7   16   16  1059   52%    -4   31%
   3 AW15     6   16   16  1053   52%    -4   29%
   4 AW13     5   17   17  1042   51%    -4   27%
   5 AW12     5   17   16  1040   51%    -4   29%
   6 AW09    -2   13   14  1598   50%    -4   31%
   7 AW06    -3    8    8  5414   50%     0   30%
   8 AW08    -4   14   13  1593   50%    -4   30%
   9 AW07    -6    8    8  5398   49%     0   31%
  10 AW11    -9   16   16  1063   49%    -4   29%
  11 AW16   -14   16   16  1137   49%    -4   30%
ResultSet-EloRating>

[Don]

I currently use it for millions of games, but they are divided into several pgns. (up to 512000 games per PGN if i am not mistaken).
Never found any limit so far, but it is slow of course as it has to parse PGNs (between half an hour and one hour to read all pgns). I would also be very interested if it could take result grids as input instead of full pgns.

The source code is freely available - I have long considered modifying it to be more scripting friendly. I wish you could do it all with simple command line arguments but instead it has it's own user interface which forces you to interact with it. I have gotten around this by building scripts which interact it with it, but I should not have to do this.

Another change I would like to make is to add an additional decimal point of resolution to the ELO ratings and error margins. The current resolution of course is usually adequate resolution for most purposes given that people rarely rate more than 100,000 games but there are times when I rate hundreds of thousands of games and I would like to see the extra digit. (Do I absolutely need the extra digit? Probably not.)

[Zach Wegner]

I currently use it for millions of games, but they are divided into several pgns. (up to 512000 games per PGN if i am not mistaken).
Never found any limit so far, but it is slow of course as it has to parse PGNs (between half an hour and one hour to read all pgns). I would also be very interested if it could take result grids as input instead of full pgns.

The source code is freely available - I have long considered modifying it to be more scripting friendly. I wish you could do it all with simple command line arguments but instead it has it's own user interface which forces you to interact with it. I have gotten around this by building scripts which interact it with it, but I should not have to do this.

Another change I would like to make is to add an additional decimal point of resolution to the ELO ratings and error margins. The current resolution of course is usually adequate resolution for most purposes given that people rarely rate more than 100,000 games but there are times when I rate hundreds of thousands of games and I would like to see the extra digit. (Do I absolutely need the extra digit? Probably not.)

I did this a long time ago, BayesElo is fully integrated into my tester. Thus no PGNs. You have a lot more control doing it this way too.

But I certainly remember loading far more than 10k games into BayesElo via PGN. Maybe there is some problem in the PGN at that point.

[Don]

I currently use it for millions of games, but they are divided into several pgns. (up to 512000 games per PGN if i am not mistaken).
Never found any limit so far, but it is slow of course as it has to parse PGNs (between half an hour and one hour to read all pgns). I would also be very interested if it could take result grids as input instead of full pgns.

The source code is freely available - I have long considered modifying it to be more scripting friendly. I wish you could do it all with simple command line arguments but instead it has it's own user interface which forces you to interact with it. I have gotten around this by building scripts which interact it with it, but I should not have to do this.

Another change I would like to make is to add an additional decimal point of resolution to the ELO ratings and error margins. The current resolution of course is usually adequate resolution for most purposes given that people rarely rate more than 100,000 games but there are times when I rate hundreds of thousands of games and I would like to see the extra digit. (Do I absolutely need the extra digit? Probably not.)

I did this a long time ago, BayesElo is fully integrated into my tester. Thus no PGNs. You have a lot more control doing it this way too.

But I certainly remember loading far more than 10k games into BayesElo via PGN. Maybe there is some problem in the PGN at that point.

Perhaps a shortcut is to pre-parse the pgn files, removing all the stuff that is not important for rating the games. This assumes that you can do this incrementally of course - trimming these files as you go and as you collect huge samples of games.

[Daniel Shawul]

I can use elostat to process the same pgn so i dont think that is the problem. But you are right 10k is not the limit. For another pgn it did well upto 43k and failed These random problem happen for big files so I will try breaking them down..

[Daniel Shawul]

I thought of that but cutechess-cli produces the pgns (with the min option), so i have to hack that too... It seems I have to modify source code to fit my needs, because everytime a useful testing tool with own format ,like QLR, pops up. I can't nag the authors to conform . If only people are generous and share their testing tools

[bob]

Is there a limit on the max number of games?. I am doing very fast games and the games pgn builds up quite quickly and it seems bayeselo can't handle more than 10000 games. Is that so ? Also is there a way to make it work by giving it only the number of wins/losses/draws ?

Thanks

I run BayesElo with millions of games. It takes a while to load them, but I have not found any limit during my cluster testing. The last batch we ran was 1.2M games and BayesElo swallowed that and produced the usual output with zero problems.

[bob]

I currently use it for millions of games, but they are divided into several pgns. (up to 512000 games per PGN if i am not mistaken).
Never found any limit so far, but it is slow of course as it has to parse PGNs (between half an hour and one hour to read all pgns). I would also be very interested if it could take result grids as input instead of full pgns.

It only takes about 30 seconds to parse a million games on our cluster "head" node during my testing...