BanksiaGUI has been developed by Nguyen Hong Pham (Phạm Hồng Nguyên).
I can be reached by banksiagui at gmail dot com. You may use comment boxes here or somewhere on this site too.
65 thoughts on “About”
awesome
hello, i just downloaded ver 0.39 today but engine doesnt work. and when i tried the 0.38 the engine also crash. when I reset the settings, it cant detect any engine.
how to fix, please.
thanks
Which engine are you using? Can you run it in command lines mode (CMD)?
Hello,
infinite mode does not work. the engine does not play.
Hello, I have registered, but the email has not arrived to validate the account walter.paterno@gmail.com
If you could solve the problem, I would be more than grateful.
Thanks for the report. It was very helpful since I was so struggling to reproduce the issue on Windows but I could on a Linux system. I have fixed the bug, uploaded a new binary (0.36b). Please try
It’s cool. It works.
So good news, thanks 😀
this exactly my experience. I don’t know how to fix it
I have sent an email with some (Spanish) translation work , please confirm if you have received it 😀
Thank you a lot. I will publish on this website soon and it will come with new releases
Here’s a proposed solution for Banksia Gui when users are playing engines over the network. Add a time bonus for each move played of 5 secs before the actual chess clock if the engine starts ticking. That is, as an example, engine A makes a move, it sends the move over to engine B which is connected in the network in another PC, engine B clock doesn’t start right away, engine B starts to ponder over the move and after 5secs engine B clock starts ticking. Time delay bonus should help to solve network lags. Of course this solution works only betwe82 Banksia GUI.
At the moment BSG has “Add time” – given extra time to the side to move before ruling the games, somewhat is a kind of bonus period, and its main purpose is to solve the issue of network lagging. Could you try and give us some feedback? Thanks a lot.
50 move rule doesn’t apply or work in the BanksiaGUI.
Engines continue to play beyond the 50 move rule. It seems that this isn’t implemented in the adjudication settings.
Thanks a lot for the report. Unfortunately, I can’t reproduce the issue. Can you post one or some games you see that problem? Thanks again.
Feature request: MANUAL ADJUDICATION.
It would be great to have a feature to manually adjudicate a current engine game like in Chessbase Fritz GUI.
Sometimes the PV is higher than 0.20 and the position is a dead draw, but the engines will keep playing until the 50 moves rule. However, sometimes one of the engines get to clever and extends the 50 moves and the games keeps going on and on until you get 200 moves or more before it is adjudicated as a Draw.
This Manual Adjudication feature would help speed up the games played by engines by allowing the user to adjudicated in cases where no further progress can be made.
Thanks a lot for the request. I am a bit confused about “manual”. Do you mean the users will adjudicate game by game manually or just set up parameters for adjudication? For setting up parameters, users can do by click on the button “setup” next to the line “Adjudication” when creating a tournament.
Yes, user will be able to manually adjudicate the game. The rules for “adjudication” works, but it doesn’t take care in situations where the position is a dead draw, but one of the engines believe it is slightly better.
A button to allow user to decide to adjudicate the game, like a win, draw or loss, that is , the user can decide the outcome of the current game instead of allowing the game to keep going. Thus, the game would officially be finished with a result and the next game can begin and the tournament can continue.
It is an interesting idea. I added it to my TODO list. Thanks
Do you need a French translator ?
Yes, of course, and thanks a lot. Actually I need help to translate BSG’s text into French as well as other languages. If you can help, please follow the instruction from the page https://banksiagui.com/internationalize-bsg/. The help is always highly appreciated.
Sorry, I just saw that you got already. Aniway, how can we change English to French in the GUI ? Thanks a lot, cheers !!
Dear Nguyen,
Thank you for the very cool GUI.
the only annoying issue , I am facing that BSG fails to play the opening pgn in sequences or even to shuffle the opening lines when you lock a player to play always white or black in Gauntlet tournament .
BSG simply keeps repeating the same opening over and over.
Could you please fix this issue
Thanks a lot for the report. It is a bug. I will fix it for the next release.
if you can, please add this very important feature in Tournament
For some reason the PGN files of the games of each tournament are not compatible with Ordo. We Use Ordo to evaluate the strength of engines when testing lc0 and the output always says: ERROR: Input file contains no games. I’ve tried with different TC such as (nodes/move or time/move) to no avail.
Thanks a lot for the report!
I think Ordo is quite strictly to parse PGN files and easily gives up if there is any extra thing. In PGN files created by BSG, there is an extra tab character (code 09, noted as \t) between tag White, Black and their values (i.e, [White \t”Stockfish”]). It is allowed by PGN standard (I think the standard did not mention about extra white characters as tab one). Almost all software accepts that but Ordo.
You may temporarily remove that character (use a text editor, search and replace) or wait for me for the next release. I will remove them anyway.
at the end of each game, please add information like these, so that we can easily find out for what reason this game is over
Thanks a lot for the report. The bug is fixed for the next release.
I really appreciate the improvements.
-Draw Adjudication was implemented VERY WELL
-Different Time Controls (Multi Timers) for Tournament has been added (This is an amazing feature that is not available in all GUIs that I have ever used)
Now there are some problems I found…
1) Multi Timers don’t work well.
What I want is that Brainfish always plays with Time Control 200000 nodes for any side (white or black) and Stockfish always plays with Time Control 1 + 0.5 for any side (white or black).
I set the Tournament like this…
Game 1
White: Brainfish play with TC 200000 nodes
Black: Stockfish play with TC 1 + 0.5
game 2
White: Stockfish play with TC 1 + 0.5
Black: Brainfish play with TC 200000 nodes
But the Tournament give different results like this …
Game 1
White: Brainfish play with TC 200000 nodes
Black: Stockfish play with TC 1 + 0.5
game 2
White: Stockfish play with TC 200000 nodes
Black: Brainfish play with TC 1 + 0.5
2) Starting Position can’t be loaded. In the previous version, Starting Positions could be loaded, now it can’t (http://prntscr.com/qi4dot)
3) I do not find the analysis feature which you said was added
4) still can’t open PGN File with many games, only PGN with one game can
other problems if I find it, I will report it here
Thanks,
please add “Analysis Button” and “Move Now Button” to the Engine Info Section like this => http://prntscr.com/qi9kkl
in this way we can immediately analyze without having to bother making settings in the “Game Analysis” menu
Thanks a lot for your report
1) It is a bug and will be fixed for the next release
2) Oops, I am going to fix now. For the current version, can you use a Polyglot book instead?
3) It is in menu->File->Analysis game. Thanks for the suggestion. I will add those buttons somewhere, perhaps, to toolbars but not in engine’s bar since it used for engine’s buttons, sometimes there are several ones
4) That feature is still in my TODO list and will be implemented within a few weeks
it seems like that “the Draw Adjudication” was not implemented, I was wrong before because I only saw data from one game
if you implement the Draw Adjudication, the game actually draw earlier, at move 86 Bc1… so save time
if you doubt about the Draw Adjudiction which I suggested before, then please just add it in the Tournament as an “Optional option”
-how to load starting positions in sequence (from the first to the last)?
=> https://prnt.sc/qfedj4
I’ve tried many times with different parameters, but it’s always not sequential.
-Users also can’t play with Time control (time/all moves + increment), this is only available in Tournaments.
So we cant play Blitz 3m+2s or other Blitz Time Control
Thanks for the report. There is a bug since random seed may affect each other. Will be fixed soon
A little thing was missed at the end of my comment …
A game longer than 50 moves / 100 ply, and both engines have an evaluation between +/- 0.09 cp for 10 consecutive moves with no pawn moves or captures, then the game is definitely Draw, so it’s a waste of time if the game still continue. Therefore, including this feature is also very important.
…In Cute Chess 1.8 mn/s at starting posisiton, but in Banksia only 1.5 mn/s, it’s equal to 16% decrease in performance.
-I mean Tournament Results in PGN format.
-A score of -500 cp for 5 consecutive move will definitely lose, so it’s a waste of time if the game still continue. Therefore it is very important to include this feature (but as an optional setting and not as a default, if user want to use this feature then it’s their own choice).
-A game with more than 50 moves / 100 plies, and both engines have an evaluation [-9,9] cp for 10 consecutive moves, then the game is definitely Draw, so it’s a waste of time if the game still continue. Therefore, including this feature is also very important.
Thanks,
Thanks for the info.
– If you run only a match of two engines their speeds on any chess GUIs should be similar. There are some factors may affect their displaying speeds: a) they use different parameters such as cores, mem/hash b) timer is not guaranty to measure the differences for just a few games – you may need to run many more games to compare c) BSG and Cute chess may use different methods to measure speed. One easier method to compare speed, let engines be limited by depth or nodes then compare their spent time periods.
– I don’t know how to store tournament results in PGN format. BSG just store game results (W/D/L) only. BSG can also write down tournament results in a text file too. If you can, please give me details on saving tour results into PGN.
– BSG can adjudicate a game by Resign score. If you set it to 500 cp, BSG will set a side be lost the game if its scores under -500 cp for 3 consecutive moves (I will change it to 5 if you think it is safer).
– Thanks for the idea about drawing scores. I will implement soon
– I already know how to get the tournament results saved in the PGN format (Maybe in the next release it can be implemented as a default setting), I made it like this in the Tournament Settings => http://prntscr.com/qgawek
-glad to hear the BSG Team will implement this drawing score
Thanks,
Thanks for the test and other info/suggestions/
For the test, the nodes are still different. Please use only 1 core for testing and make sure hash size the same too. If you can, run that engine in a terminal/console and control it manually to compare speeds with other ones.
instead of using Deepth, I choose to use Nodes because it seems more accurate, because the same Deepth can produce different nodes. In this test I used the same Nodes, the same Hash & the same Core for both GUIs.
As you can see, there is a 131 ms difference to reach the same number of Nodes.
This means, if the number of Nodes that must be reached is more, then the Time Difference will definitely be more too.
I’m not a programmer, so I don’t know what caused it, but I think something is wrong.
Or maybe you can install Cute Chess 1.0.0 or Arena 3.5.1 and do the test yourself, so we can find out if this only happen on my computer or on your computer too.
This is serious, because the impact is that engine performance can decrease and test results can be inaccurate.
IMO, the difference is not much. If you still want to find the reason, please take care that the paras you set for the test are still different (I suspect hash size different). If they are matched, all engines’ printing out must be the same (except time period). In your images, clearly depths, pv strings… are different.
-What did you said about the 3 consecutive moves in Resign Adjudication at BSG (as you can see here: http://prntscr.com/qgbi8o), actually it is only 2 consecutive moves (note that I set the resign score at -700 cp , not -500), because in “Resign Adjudication” we only count moves of the losing players, in contrast to “Draw Adjudication” where we count moves of both engines.
For example, in this game (http://prntscr.com/qgbi8o) where white loses, then the moves that are counted sequentially are only white moves, and black is not counted, so we say -500 cp for 2 consecutive moves, not 3 consecutive moves.
Now look at this => http://prntscr.com/qgbh16 … this is an example of Resign Adjudication with 5 consecutive moves in Cute Chess (note that I set the resign score to -500), you can’t say it as 10 consecutive moves.
As you can see, the game draw when the evaluations of the two engines are [-9,9] cp for 10 consecutive moves with NO PAWN MOVES or CAPTURES.
The details are…
10 consecutive ply for white and 10 consecutive ply for black
Total = 20 consecutive ply = 10 consecutive moves
…Why did I suggest this Draw Adjudication? Because Chess.com uses it at CCC (https://www.chess.com/computer-chess-championship)… and this is very effective to avoid boring draw games of engines, and save time.
Thanks for the report. I am going to check and fix
-Please Use Bayeselo to generate ELO Diff, CCRL use Bayeselo.
-Why is the CPU Speed in this GUI lower than other GUIs like Arena & Cute Chess?
for example: Single core can running 1.9 mn/s at starting position in Cute Chess, but in Banksia only 1.5 mn/s at Starting Posisiton.
-Where can I find the Tournament Results?
-in the Adjudication section in Tournament, please add these features (Draw Adjudication and Tablebase Adjudication) like in Cute Chess => http://prntscr.com/qez4fw
1) We are struggling with the algorithms of Elo. Any change will take time
2) Can you turn off all showing boards in tournaments then check again? As I guess, BSG allows showing more boards thus they may eat computing powers. In the next version, we allow to turn off animation which may help a bit
3) File -> Windows -> Dock -> Tournament Result. It is a panel with few tabs, each tab is a table about results
4) BSG can adjudicate by: a) Syzygy b) game length c) resign score. IMO, they are almost all popular methods. Do you want anything special?
Please Add these feature bellow, so we can play tournament in different time control,
For example:
White: Stockfish Bullet
Black: Stockfish Blitz
White Time Control: 1m+1s
Black Time Control: 3m+2s
The purpose is, I want to know how is the difference in ELO Rating between Bullet vs Blitz, or beetwen Bullet vs Standard, or Blitz vs Standard.
or:
White Time Control: 1m+1s
Black Time Control: deepth 20
or:
White Time Control: 3m+2s
Black Time Control: 5000000 nodes
Please add these feature in Tournament.
Thanks,
Thanks for the suggestion. Sure, we will do (we are implementing a bit more flexible way) 🙂
Hello! I was wondering if you planned to add uneven time controls to the GUI? For example: SF vs Leela, SF gets 30+5 TC and Leela gets 60+5 TC. This would be very useful for fixed node testing without having to use a command line. Thanks! 🙂
I have started working on that function already and may complete within 2 weeks
How do I run it with lc0 on a Mac OS X (Catalina)? It’s installed and runs but the help blurb from ./banksia –help shows only windows-related. Same for the json examples.
Hello Nguyên,
3) do you mean to display a move (two half moves) in a line? —-> YES
It’s easiest to read.
4)you may change notations (open Settings dialog and General tab —-> OK, i give a try.
Thanks for your reply
Hafni
Hello Nguyên,
I like your GUI but some things, IMHO, should be enhanced:
1) The font of panel “MOVE” are too tiny, it’s difficult to read!
2) Allow the panels’ background color to be customized
3) Display the moves 1 for line. e.g. 1. e2e4 e6e5 it’s more clean and readable.
2. d2d4 g8f6 etc…
4) Allow choice between style e4 or e2e4
As you can see it’s all about aesthetics but make the GUI more beautiful.
Best regards from Spain
Thank you a lot for suggestions. I have added them to my TODO list to check and work later.
For clarifying:
3) do you mean to display a move (two half moves) in a line?
4) you may change notations (open Settings dialog and General tab)
65 thoughts on “About”
awesome
hello, i just downloaded ver 0.39 today but engine doesnt work. and when i tried the 0.38 the engine also crash. when I reset the settings, it cant detect any engine.
how to fix, please.
thanks
Which engine are you using? Can you run it in command lines mode (CMD)?
Hello,
infinite mode does not work. the engine does not play.
Hello, I have registered, but the email has not arrived to validate the account walter.paterno@gmail.com
If you could solve the problem, I would be more than grateful.
Hello Nguyên,
I just translated BanksiaGUI into Simplified Chinese and tested it, here’s the link:
https://github.com/fluffyfoxxo/BanksiaGUI-Translation/blob/master/zh_CN/bsg_zh.ts
Wow, thank you very much for the contribution! It will be released soon with the new version 🙂
your new version 0.36 doesn’t manage to launch.
in manjaro kde the terminal says it: segmentation error
Thanks a lot for the report, I start working to reproduce the problem.
I can’t reproduce the issue after trying on several computers, thus it may take more time. Please use the previous version as a temporary solution.
Thank you, I’m attaching a screenshot if it can be useful
https://www.zupimages.net/up/20/33/cdku.png
Thanks for the report. It was very helpful since I was so struggling to reproduce the issue on Windows but I could on a Linux system. I have fixed the bug, uploaded a new binary (0.36b). Please try
It’s cool. It works.
So good news, thanks 😀
this exactly my experience. I don’t know how to fix it
I have sent an email with some (Spanish) translation work , please confirm if you have received it 😀
Thank you a lot. I will publish on this website soon and it will come with new releases
Here’s a proposed solution for Banksia Gui when users are playing engines over the network. Add a time bonus for each move played of 5 secs before the actual chess clock if the engine starts ticking. That is, as an example, engine A makes a move, it sends the move over to engine B which is connected in the network in another PC, engine B clock doesn’t start right away, engine B starts to ponder over the move and after 5secs engine B clock starts ticking. Time delay bonus should help to solve network lags. Of course this solution works only betwe82 Banksia GUI.
At the moment BSG has “Add time” – given extra time to the side to move before ruling the games, somewhat is a kind of bonus period, and its main purpose is to solve the issue of network lagging. Could you try and give us some feedback? Thanks a lot.
50 move rule doesn’t apply or work in the BanksiaGUI.
Engines continue to play beyond the 50 move rule. It seems that this isn’t implemented in the adjudication settings.
Thanks a lot for the report. Unfortunately, I can’t reproduce the issue. Can you post one or some games you see that problem? Thanks again.
Feature request: MANUAL ADJUDICATION.
It would be great to have a feature to manually adjudicate a current engine game like in Chessbase Fritz GUI.
Sometimes the PV is higher than 0.20 and the position is a dead draw, but the engines will keep playing until the 50 moves rule. However, sometimes one of the engines get to clever and extends the 50 moves and the games keeps going on and on until you get 200 moves or more before it is adjudicated as a Draw.
This Manual Adjudication feature would help speed up the games played by engines by allowing the user to adjudicated in cases where no further progress can be made.
Thanks a lot for the request. I am a bit confused about “manual”. Do you mean the users will adjudicate game by game manually or just set up parameters for adjudication? For setting up parameters, users can do by click on the button “setup” next to the line “Adjudication” when creating a tournament.
Yes, user will be able to manually adjudicate the game. The rules for “adjudication” works, but it doesn’t take care in situations where the position is a dead draw, but one of the engines believe it is slightly better.
A button to allow user to decide to adjudicate the game, like a win, draw or loss, that is , the user can decide the outcome of the current game instead of allowing the game to keep going. Thus, the game would officially be finished with a result and the next game can begin and the tournament can continue.
It is an interesting idea. I added it to my TODO list. Thanks
Do you need a French translator ?
Yes, of course, and thanks a lot. Actually I need help to translate BSG’s text into French as well as other languages. If you can help, please follow the instruction from the page https://banksiagui.com/internationalize-bsg/. The help is always highly appreciated.
Sorry, I just saw that you got already. Aniway, how can we change English to French in the GUI ? Thanks a lot, cheers !!
Just download the file https://banksiagui.com/dl/bsg_en.ts.zip then translate text between tags then send back to us (via email). Instruction at https://banksiagui.com/internationalize-bsg/. Thanks in advance.
Dear Nguyen,
Thank you for the very cool GUI.
the only annoying issue , I am facing that BSG fails to play the opening pgn in sequences or even to shuffle the opening lines when you lock a player to play always white or black in Gauntlet tournament .
BSG simply keeps repeating the same opening over and over.
Could you please fix this issue
Thanks a lot for the report. It is a bug. I will fix it for the next release.
if you can, please add this very important feature in Tournament
Wait between games => http://prntscr.com/qlk9wu
thanks,
so the computer get
For some reason the PGN files of the games of each tournament are not compatible with Ordo. We Use Ordo to evaluate the strength of engines when testing lc0 and the output always says: ERROR: Input file contains no games. I’ve tried with different TC such as (nodes/move or time/move) to no avail.
in the Tournament section, set it like this => http://prntscr.com/qlbiar
then you get a PGN file of the games
Thanks a lot for the report!
I think Ordo is quite strictly to parse PGN files and easily gives up if there is any extra thing. In PGN files created by BSG, there is an extra tab character (code 09, noted as \t) between tag White, Black and their values (i.e, [White \t”Stockfish”]). It is allowed by PGN standard (I think the standard did not mention about extra white characters as tab one). Almost all software accepts that but Ordo.
You may temporarily remove that character (use a text editor, search and replace) or wait for me for the next release. I will remove them anyway.
at the end of each game, please add information like these, so that we can easily find out for what reason this game is over
Draw by Adjudication => http://prntscr.com/qjsxj9
—you can change this one like this “Draw by adjudication: BSG draw rule”—
Draw by 3-fold repetition => http://prntscr.com/qjsprz
Draw by adjudication: SyzygyTB => http://prntscr.com/qjsq6q
Draw by fifty moves rule => http://prntscr.com/qjsqhi
White wins by adjudication => http://prntscr.com/qjsqyq
—you can change this one like this “White wins by adjudication: BSG resign rule”—
Black wins by adjudication => http://prntscr.com/qjsqrh
—you can change this one like this “Black wins by adjudication: BSG resign rule”—
White wins by adjudication: SyzygyTB => http://prntscr.com/qjsssm
Black wins by adjudication: SyzygyTB => http://prntscr.com/qjsrse
Thanks a lot for the suggestion => my TODO list now
i found this problem in Tournament result
=> http://prntscr.com/qj41x1
Thanks a lot for the report. The bug is fixed for the next release.
I really appreciate the improvements.
-Draw Adjudication was implemented VERY WELL
-Different Time Controls (Multi Timers) for Tournament has been added (This is an amazing feature that is not available in all GUIs that I have ever used)
Now there are some problems I found…
1) Multi Timers don’t work well.
As you can see here => http://prntscr.com/qi4ayl
What I want is that Brainfish always plays with Time Control 200000 nodes for any side (white or black) and Stockfish always plays with Time Control 1 + 0.5 for any side (white or black).
I set the Tournament like this…
Game 1
White: Brainfish play with TC 200000 nodes
Black: Stockfish play with TC 1 + 0.5
game 2
White: Stockfish play with TC 1 + 0.5
Black: Brainfish play with TC 200000 nodes
But the Tournament give different results like this …
Game 1
White: Brainfish play with TC 200000 nodes
Black: Stockfish play with TC 1 + 0.5
game 2
White: Stockfish play with TC 200000 nodes
Black: Brainfish play with TC 1 + 0.5
2) Starting Position can’t be loaded. In the previous version, Starting Positions could be loaded, now it can’t (http://prntscr.com/qi4dot)
3) I do not find the analysis feature which you said was added
4) still can’t open PGN File with many games, only PGN with one game can
other problems if I find it, I will report it here
Thanks,
please add “Analysis Button” and “Move Now Button” to the Engine Info Section like this => http://prntscr.com/qi9kkl
Arena 3.5.1 implemented these very well => http://prntscr.com/qi9iy5
in this way we can immediately analyze without having to bother making settings in the “Game Analysis” menu
Thanks a lot for your report
1) It is a bug and will be fixed for the next release
2) Oops, I am going to fix now. For the current version, can you use a Polyglot book instead?
3) It is in menu->File->Analysis game. Thanks for the suggestion. I will add those buttons somewhere, perhaps, to toolbars but not in engine’s bar since it used for engine’s buttons, sometimes there are several ones
4) That feature is still in my TODO list and will be implemented within a few weeks
it seems like that “the Draw Adjudication” was not implemented, I was wrong before because I only saw data from one game
this game (http://prntscr.com/qjsa9b) finally draw after 126 moves / 252 ply was played
if you implement the Draw Adjudication, the game actually draw earlier, at move 86 Bc1… so save time
if you doubt about the Draw Adjudiction which I suggested before, then please just add it in the Tournament as an “Optional option”
-how to load starting positions in sequence (from the first to the last)?
=> https://prnt.sc/qfedj4
I’ve tried many times with different parameters, but it’s always not sequential.
-Users also can’t play with Time control (time/all moves + increment), this is only available in Tournaments.
So we cant play Blitz 3m+2s or other Blitz Time Control
Thanks for the report. There is a bug since random seed may affect each other. Will be fixed soon
A little thing was missed at the end of my comment …
A game longer than 50 moves / 100 ply, and both engines have an evaluation between +/- 0.09 cp for 10 consecutive moves with no pawn moves or captures, then the game is definitely Draw, so it’s a waste of time if the game still continue. Therefore, including this feature is also very important.
Thanks
-This what i mean:
Speed on Banksia => http://prntscr.com/qf1avt
Speed on Cute Chess => https://prnt.sc/qf1a0q
…In Cute Chess 1.8 mn/s at starting posisiton, but in Banksia only 1.5 mn/s, it’s equal to 16% decrease in performance.
-I mean Tournament Results in PGN format.
-A score of -500 cp for 5 consecutive move will definitely lose, so it’s a waste of time if the game still continue. Therefore it is very important to include this feature (but as an optional setting and not as a default, if user want to use this feature then it’s their own choice).
-A game with more than 50 moves / 100 plies, and both engines have an evaluation [-9,9] cp for 10 consecutive moves, then the game is definitely Draw, so it’s a waste of time if the game still continue. Therefore, including this feature is also very important.
Thanks,
Thanks for the info.
– If you run only a match of two engines their speeds on any chess GUIs should be similar. There are some factors may affect their displaying speeds: a) they use different parameters such as cores, mem/hash b) timer is not guaranty to measure the differences for just a few games – you may need to run many more games to compare c) BSG and Cute chess may use different methods to measure speed. One easier method to compare speed, let engines be limited by depth or nodes then compare their spent time periods.
– I don’t know how to store tournament results in PGN format. BSG just store game results (W/D/L) only. BSG can also write down tournament results in a text file too. If you can, please give me details on saving tour results into PGN.
– BSG can adjudicate a game by Resign score. If you set it to 500 cp, BSG will set a side be lost the game if its scores under -500 cp for 3 consecutive moves (I will change it to 5 if you think it is safer).
– Thanks for the idea about drawing scores. I will implement soon
– Spent time periode to reach deepth 20 on BSG, Cute Chess 1.0.0 & Arena 3.5.1
BSG => http://prntscr.com/qgamqj
Cute Chess 1.0.0 => http://prntscr.com/qgangs
Arena 3.5.1 => http://prntscr.com/qgaovu
Engine: Brainfish 221119 64 POPCNT
BSG
deepth: 20
Nodes: 11338000
time: 7 seconds
speed: 11338000/7 = 1619Kn/s
Cute Chess 1.0.0
deepth: 20
Nodes: 10267878
time: 5.6 seconds
speed: 10267878/5.6 = 1833Kn/s
Arena 3.5.1
deepth: 20
Nodes: 9626743
time: 4.964 seconds
speed: 9626743/4.964 = 1939Kn/s
…hopefully in the future it can be fixed
– I already know how to get the tournament results saved in the PGN format (Maybe in the next release it can be implemented as a default setting), I made it like this in the Tournament Settings => http://prntscr.com/qgawek
…and the results are like this:
– http://prntscr.com/qgb2tf
-http://prntscr.com/qgb360
-glad to hear the BSG Team will implement this drawing score
Thanks,
Thanks for the test and other info/suggestions/
For the test, the nodes are still different. Please use only 1 core for testing and make sure hash size the same too. If you can, run that engine in a terminal/console and control it manually to compare speeds with other ones.
instead of using Deepth, I choose to use Nodes because it seems more accurate, because the same Deepth can produce different nodes. In this test I used the same Nodes, the same Hash & the same Core for both GUIs.
And here the results…
BSG: http://prntscr.com/qgyry9
Cute Chess: http://prntscr.com/qgytds
As you can see, there is a 131 ms difference to reach the same number of Nodes.
This means, if the number of Nodes that must be reached is more, then the Time Difference will definitely be more too.
I’m not a programmer, so I don’t know what caused it, but I think something is wrong.
Or maybe you can install Cute Chess 1.0.0 or Arena 3.5.1 and do the test yourself, so we can find out if this only happen on my computer or on your computer too.
This is serious, because the impact is that engine performance can decrease and test results can be inaccurate.
IMO, the difference is not much. If you still want to find the reason, please take care that the paras you set for the test are still different (I suspect hash size different). If they are matched, all engines’ printing out must be the same (except time period). In your images, clearly depths, pv strings… are different.
-What did you said about the 3 consecutive moves in Resign Adjudication at BSG (as you can see here: http://prntscr.com/qgbi8o), actually it is only 2 consecutive moves (note that I set the resign score at -700 cp , not -500), because in “Resign Adjudication” we only count moves of the losing players, in contrast to “Draw Adjudication” where we count moves of both engines.
For example, in this game (http://prntscr.com/qgbi8o) where white loses, then the moves that are counted sequentially are only white moves, and black is not counted, so we say -500 cp for 2 consecutive moves, not 3 consecutive moves.
Now look at this => http://prntscr.com/qgbh16 … this is an example of Resign Adjudication with 5 consecutive moves in Cute Chess (note that I set the resign score to -500), you can’t say it as 10 consecutive moves.
-About the Draw Adjudication I mean, see here => https://prnt.sc/qgbkww
As you can see, the game draw when the evaluations of the two engines are [-9,9] cp for 10 consecutive moves with NO PAWN MOVES or CAPTURES.
The details are…
10 consecutive ply for white and 10 consecutive ply for black
Total = 20 consecutive ply = 10 consecutive moves
…Why did I suggest this Draw Adjudication? Because Chess.com uses it at CCC (https://www.chess.com/computer-chess-championship)… and this is very effective to avoid boring draw games of engines, and save time.
Thanks for the report. I am going to check and fix
-Please Use Bayeselo to generate ELO Diff, CCRL use Bayeselo.
-Why is the CPU Speed in this GUI lower than other GUIs like Arena & Cute Chess?
for example: Single core can running 1.9 mn/s at starting position in Cute Chess, but in Banksia only 1.5 mn/s at Starting Posisiton.
-Where can I find the Tournament Results?
-in the Adjudication section in Tournament, please add these features (Draw Adjudication and Tablebase Adjudication) like in Cute Chess => http://prntscr.com/qez4fw
1) We are struggling with the algorithms of Elo. Any change will take time
2) Can you turn off all showing boards in tournaments then check again? As I guess, BSG allows showing more boards thus they may eat computing powers. In the next version, we allow to turn off animation which may help a bit
3) File -> Windows -> Dock -> Tournament Result. It is a panel with few tabs, each tab is a table about results
4) BSG can adjudicate by: a) Syzygy b) game length c) resign score. IMO, they are almost all popular methods. Do you want anything special?
Please Add these feature bellow, so we can play tournament in different time control,
For example:
White: Stockfish Bullet
Black: Stockfish Blitz
White Time Control: 1m+1s
Black Time Control: 3m+2s
The purpose is, I want to know how is the difference in ELO Rating between Bullet vs Blitz, or beetwen Bullet vs Standard, or Blitz vs Standard.
or:
White Time Control: 1m+1s
Black Time Control: deepth 20
or:
White Time Control: 3m+2s
Black Time Control: 5000000 nodes
Please add these feature in Tournament.
Thanks,
Thanks for the suggestion. Sure, we will do (we are implementing a bit more flexible way) 🙂
Hello! I was wondering if you planned to add uneven time controls to the GUI? For example: SF vs Leela, SF gets 30+5 TC and Leela gets 60+5 TC. This would be very useful for fixed node testing without having to use a command line. Thanks! 🙂
I have started working on that function already and may complete within 2 weeks
How do I run it with lc0 on a Mac OS X (Catalina)? It’s installed and runs but the help blurb from ./banksia –help shows only windows-related. Same for the json examples.
Hello Nguyên,
3) do you mean to display a move (two half moves) in a line? —-> YES
It’s easiest to read.
4)you may change notations (open Settings dialog and General tab —-> OK, i give a try.
Thanks for your reply
Hafni
Hello Nguyên,
I like your GUI but some things, IMHO, should be enhanced:
1) The font of panel “MOVE” are too tiny, it’s difficult to read!
2) Allow the panels’ background color to be customized
3) Display the moves 1 for line. e.g. 1. e2e4 e6e5 it’s more clean and readable.
2. d2d4 g8f6 etc…
4) Allow choice between style e4 or e2e4
As you can see it’s all about aesthetics but make the GUI more beautiful.
Best regards from Spain
Thank you a lot for suggestions. I have added them to my TODO list to check and work later.
For clarifying:
3) do you mean to display a move (two half moves) in a line?
4) you may change notations (open Settings dialog and General tab)
Comments are closed.