Elements of the ULTIMATE Chess GUI?

[voffka]

Hello everyone,

WinBoard/XBoard support time odds as well, of course. Just not different TC modes. Not because it would be difficult to program, but because it seems completely useless, or even worth discouraging.

It's interesting to test with different TCs.
Even if it has no real use. Just different numbers.
It's like "playing" an idle clicker game. Total waste of time and electricity but sometimes you just have that urge to know what kind of time advantage Igel needs to beat Stockfish.

Note that HGM spoke off different TC 'modes' not plain time odds like you seem to suggest for your tests.
I have to admit though that I ran into some problems with time odds in WB.

I hope in a year or two, Igel will beat Stockfish on the same time control and the same hardware Currently in all my tests the elo difference between them is infinity.

[Guenther]

WB.ini entries (only first part)

Code: Select all

"Booot_62-64" /fd=C:\Engines\UCI\Booot_62-64 -fn Booot_62-64 /fUCCI
"Booot_62-64#2" /fd=C:\Engines\UCI\Booot_62-64#2 -fn Booot_62-64#2 /fUCCI -firstTimeOdds 2
"Booot_62-64#8" /fd=C:\Engines\UCI\Booot_62-64#8 -fn Booot_62-64#8 /fUCCI -firstTimeOdds 8
...
"Fizbo_19-64" /fcp "Fizbo_19-64" /fd=C:\Engines\UCI\Fizbo_19-64 -fn Fizbo_19-64 /fUCCI
"Fizbo_19-64#2" /fcp "Fizbo_19-64#2" /fd=C:\Engines\UCI\Fizbo_19-64#2 -fn Fizbo_19-64#2 /fUCCI -firstTimeOdds 2
"Fizbo_19-64#8" /fcp "Fizbo_19-64#8" /fd=C:\Engines\UCI\Fizbo_19-64#8 -fn Fizbo_19-64#8 /fUCCI -firstTimeOdds 8

Code: Select all

-participants {Booot_62-64
Booot_62-64#2
Booot_62-64#8
Fizbo_19-64
Fizbo_19-64#2
Fizbo_19-64#8
...

Umm, it seems I never noticed your last post in that thread. :oops:

The problem you encounter there is because of wrong use of the time-odds feature. When you want the time odds to follow an engine in a tournament, you have to install that engine with 'Extra WinBoard options' (in the Load Engine dialog) /firstTimeOdds=N . The time odds entries in the Time Control Options dialog are for first and second engine, but in a tournament (or simple match) you would never know for a given pairing which engine is first and which is second, and there might also be games in the tourney between engines that have no time odds at all.

I don't understand this. Isn't this exactly what my WB.ini excerpt shows? (ofc I have left out the 'second' part, as it is just mirrored)

[WinPooh]

Support of ultra-fast time controls (like CUTECLI, unlike Arena). Non-equal time controls (for example, engine A: nodes per move, engine B: fixed depth).

[WinPooh]

I hope in a year or two, Igel will beat Stockfish on the same time control and the same hardware Currently in all my tests the elo difference between them is infinity.

Stockfish is obsolete, lc0 seems to be a more worthy goal

[Dann Corbit]

I hope in a year or two, Igel will beat Stockfish on the same time control and the same hardware Currently in all my tests the elo difference between them is infinity.

Stockfish is obsolete, lc0 seems to be a more worthy goal

Reminds me of the Mark Twain quote:
"The reports of my death are greatly exaggerated"

Everyone sees the progress of LC0.
But SF marches forward too.

Sure, LC0 is playing famously after a very short development effort, compared to other engines.
But consider also that high end GPUs have about 10 trillion FLOPS of compute power in single precision and more in the tiny float 16 and smaller formats.
https://en.wikipedia.org/wiki/Nvidia_Tesla
Not sure that those Google TPUs can do, but I guess they are similar to the NVIDIA parts.

The new AMD EPYC 7601 in dual-CPU were pushing 1242 GFLOPs @ 2.2GHz base (2.7GHz boost) {so that is 1.2 TFLOPs}
https://www.tweaktown.com/news/58180/am ... index.html
So a single GPU has ten times more horsepower than a super-duper pair of high end, bleeding edge CPUs.

Eventually, I guess that a hybrid approach will win out. (Maybe 512 CPU cores and 20K GPU threads).
I bet it would be smarter than the average bear.

[Jonathan003]

I think the ultimate Chess GUI already exist.
It cannot be that we are 2021 and there is still noting better than Chessbase.
These programmers just keep their ideas and programs for them self.
Stefan Renzewitz is one of the best chess programmers for chess GUI to help humans to become better chess players. But I’m still waiting for Chess Position Trainer 6.

[Cornfed]

I think the ultimate Chess GUI already exist.
It cannot be that we are 2021 and there is still noting better than Chessbase.
These programmers just keep their ideas and programs for them self.
Stefan Renzewitz is one of the best chess programmers for chess GUI to help humans to become better chess players. But I’m still waiting for Chess Position Trainer 6.

It's been, what...3 years since he posted CPT 6 would be out 'soon'??
Love CPT...but he really turned me off.
Not just the promises he kept finding excuses for not meeting...but that damn 'Yearly Subscription' service which you got nothing for in the end. He is either (insert your choice) or a bad businessman. The latter I think...

[phhnguyen]

...
What is it that you wish to do that you can't do with any of the current GUIs?

I, for one, want fun. Chess is a game and some people have taken it so seriously that it's a chore for them to run tournaments just to get some numbers that spell some elo difference, that could have been gotten by someone else instead of them, and they don't even realize it.

For some fun I'd want to see the realization of my idea:

Mixed engines.

The problem that we currently have is that chess entities are fixed. Stockfish is always Stockfish and Komodo is always Komodo.

Imagine that Houdini starts the game and that it switches to Komodo after a 0.40 centi-pawn advantage is reached or to Stockfish if you go below -0.20 for defense, or whatever.

Imagine an engine that becomes another after every 20 moves of the game.

An engine that only appears and makes a move after some other engine has made a capture, or was put in check.

This would be really fun to watch in games, as the super-solid chess entity that started a game may suddenly go bonkers and win with a surprise king attack that the original never saw coming, or you can put to use that super-speculative personality of your engine because once it reaches that winning position it will not give it away because it no longer keeps playing crazy but switches to an engine that can finish the game but wouldn't have reached that position in the first place.

I wouldn't be surprised if some combination of this concept made appear a chess entity much stronger than the Komodos, Stockfishes and Houdinis of today, not only by fusing their strength, but perhaps throwing it a Hanibbal, Andcacs or Fizbo whenever it's pertinent and they'd play a striking move that the top engines would prune.

Is this an element of the ULTIMATE Chess GUI?

Doubtful, but I think users have to think what they want that they don't have already and ask for it instead of bloating something with stuff they already have just so they don't have to open one more GUI to do the job.

I am very interested in that idea and have been managing to implement it for the next release of BanksiaGUI. Thanks

[CMCanavessi]

I'm still looking for a low-latency GUI (Arena is an example of the opposite) that would have the same flexibility that Arena has regarding tournaments, that is that you can save them in a file that would contain all the info, so you can load it whenever you want, you can play some games, then stop, load another tournament, play some games, stop, power off the pc, turn it on again, resume the 1st tournament, edit the list of participants and it would automatically adjust the schedule, etc etc. So far all the GUIs that I've found will just have the option of running 1 tournament, but when you want to create another, the previous one will disappear and you'll lose everything.

Another thing that would be great, is having an option to automatically re-run (replay) games that finish in an abnormal way (be it time forfeits, illegal moves, engine crashes, etc) with some sort of sub-option to save the games that finished in an abnormal way in a different pgn (or even discard them), so the main tournament pgn only contains "good" games.
I don't mind editing .txt files, I do it with Arena all the time.

[JohnW]

Support for electronic chessboards like DGT and Certabo without the need for the software that currently runs between gui and board.
A good database management system to save games similar to the Chessbase software that comes with Fritz but not proprietary.
Ability to annotate a game similar to what Fritz can do but better so that it can review my games, tell me where I made mistakes and what I should have done.