Fide equivalent "slow blitz" rating list

[lkaufman]

Here are some data I collected for Dragon 2.61. In the frustration of all illegal moves I had to trow in Hiarcs and Shredder. Even TSCP have illegal moves.

Book used are the first 50 games in Günter's random 3M book played with switching side.


This give a performance between each Dragon Elo like this:

I'm using only matches where the score is between 20% and 80%. The performance differences can then be calculated with this simple formula:
800*(ScoreHight-ScoreLow)/games

I also did a selftest with 2374 games for each match which gave this result:

The book here are all games in Günter's book (1187*2).

In all list the selected games are marked with green (20-80%).
All games can be found here: http://polarchess.net/Filer/DragonGames.zip

To find engines I used:
https://rwbc-chess.de/download.htm
http://www.computer-chess.org/doku.php? ... nload_list
http://kirr.homeunix.org/chess/engines/ ... ollection/
Carlos777's list above
OliverBr list over bugfree engines for mac and linux
+google of course


It looks like there are some problem between 1600-1700-1800 in the elo setting of Dragon

The hard job with finding engines without problem and for wich level are done so it is easy to play more games against forregn engines to get a more accurate list if it is needed.

Using Monchester 1.0 as my anchor to human play, I get a rather different result. Dragon with elo 1240 seems to be about equal to Monchester based on your data, which would be the same at 1' + 1" or at a longer tc since neither engine searches deeper with more time. But at 5' + 3" Lichess human performance translated to 1503 FIDE, so it would seem that based on this the human will need somewhere in the Rapid range to compete with equal rated Dragon; presumably Dragon 1240 would crush human FIDE 1240 at 5' + 3" based on this data. Maybe this means that the time control needed for human/Dragon parity is somewhere between 5' + 3" and 15' + 10", for example 10' + 5".

[Guenther]

Here are some data I collected for Dragon 2.61. In the frustration of all illegal moves I had to trow in Hiarcs and Shredder. Even TSCP have illegal moves.

Book used are the first 50 games in Günter's random 3M book played with switching side.

...

To find engines I used:
https://rwbc-chess.de/download.htm

...

I guess you just forgot my chronology, because it is the one most used for looking for engine links and the only one still updated?
My download page just offers a very small fraction of programs I compiled myself (often for n/a old hardware or dev versions), or programs
which have neither a website, nor a working wayback archive anymore.

https://rwbc-chess.de/chronology.htm

[Guenther]

Here are some data I collected for Dragon 2.61. In the frustration of all illegal moves I had to trow in Hiarcs and Shredder. Even TSCP have illegal moves.

...
In all list the selected games are marked with green (20-80%).
All games can be found here: http://polarchess.net/Filer/DragonGames.zip

I looked very shortly now at a few games. Does Dragon at 1000 level really play that way?
In only 5 moves after the start position it already refused two times to take back a full piece
after it was captured? I doubt this makes much sense or feels realistic?
(Ofc I have no idea how people at 1000 'think', but I assume they know at least the material value
of the pieces)

[pgn][Event "My Tournament"]
[Site "?"]
[Date "2022.01.14"]
[Round "1"]
[White "Monchester 1.0"]
[Black "Dragon 2.6.1 1000 Default"]
[Result "1-0"]
[ECO "A45"]
[GameDuration "00:00:02"]
[GameEndTime "2022-01-14T17:23:47.268 Vest-Europa (normaltid)"]
[GameStartTime "2022-01-14T17:23:44.484 Vest-Europa (normaltid)"]
[Opening "Trompovsky attack (Ruth, Opovcensky Opening)"]
[PlyCount "111"]
[TimeControl "60+1"]

1. d4 {book} Nf6 {book} 2. Bg5 {book} d5 {book} 3. Nc3 {book} c5 {book}
4. Bxf6 {+2.28/4 0.14s} Qd6 {+1.27/1 0.001s} 5. Be5 {+11.97/4 0.13s}
Qa6 {-2.14/1 0s} 6. Bxb8 {+11.82/4 0.15s} f6 {-3.88/1 0s}[/pgn]

Edit:
BTW I am still interested in verified bugs in all engines listed in my chronology for being mentioned in the 'Details' sheet.
The illegal moves you found with TSCP (and especially which version?), are those such like mentioned here:
forum3/viewtopic.php?f=7&t=57318&hilit= ... rpromotion

TSCP fixes:

Code: Select all

                               VERSION HISTORY

Version 1.81c, 2/3/19
* Fixed the makefile so it compiles the .c files with optimizations.

Version 1.81b, 3/10/16
* Small revisions to this README file; updated my e-mail address and
  home page URL.
* Separated TSCP's source code and the Windows executable.
* Added a makefile.
* Updated parse_move in main.c to allow lower-case letters for pawn
  promotions.
* print_result claimed a threefold repetition draw after a position occurred
  four times instead of three. Fixed.

[lkaufman]

Here are some data I collected for Dragon 2.61. In the frustration of all illegal moves I had to trow in Hiarcs and Shredder. Even TSCP have illegal moves.

...
In all list the selected games are marked with green (20-80%).
All games can be found here: http://polarchess.net/Filer/DragonGames.zip

I looked very shortly now at a few games. Does Dragon at 1000 level really play that way?
In only 5 moves after the start position it already refused two times to take back a full piece
after it was captured? I doubt this makes much sense or feels realistic?
(Ofc I have no idea how people at 1000 'think', but I assume they know at least the material value
of the pieces)

[pgn][Event "My Tournament"]
[Site "?"]
[Date "2022.01.14"]
[Round "1"]
[White "Monchester 1.0"]
[Black "Dragon 2.6.1 1000 Default"]
[Result "1-0"]
[ECO "A45"]
[GameDuration "00:00:02"]
[GameEndTime "2022-01-14T17:23:47.268 Vest-Europa (normaltid)"]
[GameStartTime "2022-01-14T17:23:44.484 Vest-Europa (normaltid)"]
[Opening "Trompovsky attack (Ruth, Opovcensky Opening)"]
[PlyCount "111"]
[TimeControl "60+1"]

1. d4 {book} Nf6 {book} 2. Bg5 {book} d5 {book} 3. Nc3 {book} c5 {book}
4. Bxf6 {+2.28/4 0.14s} Qd6 {+1.27/1 0.001s} 5. Be5 {+11.97/4 0.13s}
Qa6 {-2.14/1 0s} 6. Bxb8 {+11.82/4 0.15s} f6 {-3.88/1 0s}[/pgn]

Edit:
BTW I am still interested in verified bugs in all engines listed in my chronology for being mentioned in the 'Details' sheet.
The illegal moves you found with TSCP (and especially which version?), are those such like mentioned here:
forum3/viewtopic.php?f=7&t=57318&hilit= ... rpromotion

TSCP fixes:

Code: Select all

                               VERSION HISTORY

Version 1.81c, 2/3/19
* Fixed the makefile so it compiles the .c files with optimizations.

Version 1.81b, 3/10/16
* Small revisions to this README file; updated my e-mail address and
  home page URL.
* Separated TSCP's source code and the Windows executable.
* Added a makefile.
* Updated parse_move in main.c to allow lower-case letters for pawn
  promotions.
* print_result claimed a threefold repetition draw after a position occurred
  four times instead of three. Fixed.

When Dragon Elo is set to 1000 (or lower values) it will indeed make random piece blunders sometimes, but I must admit that simply failing to recapture a piece is a nontypical human blunder (though I do see it sometimes with novice players), perhaps we can improve this.

[Ferdy]

Here are some data I collected for Dragon 2.61. In the frustration of all illegal moves I had to trow in Hiarcs and Shredder. Even TSCP have illegal moves.

What hardware was used, could you post the results of stockfish 10 bench command? I would like to estimate how close it is in CCRL Blitz hardware (Time control: From Jan 2020, equivalent to 2'+1" on an Intel i7-4770K).

[Odd Gunnar Malin]

Here are some data I collected for Dragon 2.61. In the frustration of all illegal moves I had to trow in Hiarcs and Shredder. Even TSCP have illegal moves.

What hardware was used, could you post the results of stockfish 10 bench command? I would like to estimate how close it is in CCRL Blitz hardware (Time control: From Jan 2020, equivalent to 2'+1" on an Intel i7-4770K).

Hi.
Sorry but I have a busy shedule this week and haven't any spare time, hopefully it will be better in the weekend and I can answer all the above (to Günter anf Larry).
The machine I was running this on was a laptop with: Prosessor Intel(R) Core(TM) i7-10750H CPU @ 2.60GHz, 2592 Mhz, 6 kjerne(r), 12 logiske prosessor(er)
It have 16GB ram and I did run 6 matches in paralell all with single thread and ponder off.
For Dragon this isn't important since it wil play the same whatever your cpu or timecontrol is. I see I forgot to metion that it was 1+1 games. It is in the gamefile though.
For the other engines some (most) will play better with longer time and maybe some of the engines that was reported 'stalled' or 'lost on time' would be helped with longer timecontrol. I didn't include those engines in the list. I did use CuteChess (GUI) for all games.
I see now that I should probably used some sort of adjucating since Dragon on lower level have problem with winning with a queen up. In a human-engine game the human would have resigned long before and it is there the data hopefully could be used. All I was interested in from the other engines was that they did run without any error in 1+1 matches. To use adjucating you probably need to check the engines a little more to see that they have a reliable pv line.
I don't know how to run this bench you talk about, I could look at it in the weekend. All my use of Stockfish is through the web. SF 11 on Chess24 and SF 14 on Lichess. I see that Fritz (or chessbase) did come with Stockfish 13 so it is resting on my harddrive ready to be used.

[Ferdy]

Here are some data I collected for Dragon 2.61. In the frustration of all illegal moves I had to trow in Hiarcs and Shredder. Even TSCP have illegal moves.

What hardware was used, could you post the results of stockfish 10 bench command? I would like to estimate how close it is in CCRL Blitz hardware (Time control: From Jan 2020, equivalent to 2'+1" on an Intel i7-4770K).

Hi.
Sorry but I have a busy shedule this week and haven't any spare time, hopefully it will be better in the weekend and I can answer all the above (to Günter anf Larry).
The machine I was running this on was a laptop with: Prosessor Intel(R) Core(TM) i7-10750H CPU @ 2.60GHz, 2592 Mhz, 6 kjerne(r), 12 logiske prosessor(er)
It have 16GB ram and I did run 6 matches in paralell all with single thread and ponder off.
For Dragon this isn't important since it wil play the same whatever your cpu or timecontrol is. I see I forgot to metion that it was 1+1 games. It is in the gamefile though.
For the other engines some (most) will play better with longer time and maybe some of the engines that was reported 'stalled' or 'lost on time' would be helped with longer timecontrol. I didn't include those engines in the list. I did use CuteChess (GUI) for all games.
I see now that I should probably used some sort of adjucating since Dragon on lower level have problem with winning with a queen up. In a human-engine game the human would have resigned long before and it is there the data hopefully could be used. All I was interested in from the other engines was that they did run without any error in 1+1 matches. To use adjucating you probably need to check the engines a little more to see that they have a reliable pv line.
I don't know how to run this bench you talk about, I could look at it in the weekend. All my use of Stockfish is through the web. SF 11 on Chess24 and SF 14 on Lichess. I see that Fritz (or chessbase) did come with Stockfish 13 so it is resting on my harddrive ready to be used.

Stockfish 10 is at https://www.dropbox.com/sh/75gzfgu7qo94 ... Br4r_RwhGa.
Try the bmi and popcnt and take the bench with the least total time.
Just run it and type bench.

Example:

Code: Select all

PS F:\Downloads\stockfish-10-win\Windows> ./stockfish_10_x64_popcnt.exe
Stockfish 10 64 POPCNT by T. Romstad, M. Costalba, J. Kiiski, G. Linscott
bench

...
Total time (ms) : 3569
Nodes searched  : 3939338
Nodes/second    : 1103765

[Odd Gunnar Malin]

Running from PowerShell. First run after start. If I run it several time it can vary +/- 1-2%

Code: Select all

		
stockfish_10_x64_popcnt.exe
===========================
Total time (ms) : 1619
Nodes searched  : 3939338
Nodes/second    : 2433192

stockfish_10_x64_bmi2.exe
===========================
Total time (ms) : 1597
Nodes searched  : 3939338
Nodes/second    : 2466711

stockfish_10_x64.exe
===========================
Total time (ms) : 1725
Nodes searched  : 3939338
Nodes/second    : 2283674

stockfish_10_x32.exe
===========================
Total time (ms) : 2157
Nodes searched  : 3939338
Nodes/second    : 1826304

Edit: Changed it to fastest of 5-6 run.

[Uri Blass]

Here are some data I collected for Dragon 2.61. In the frustration of all illegal moves I had to trow in Hiarcs and Shredder. Even TSCP have illegal moves.

What hardware was used, could you post the results of stockfish 10 bench command? I would like to estimate how close it is in CCRL Blitz hardware (Time control: From Jan 2020, equivalent to 2'+1" on an Intel i7-4770K).

Hi.
Sorry but I have a busy shedule this week and haven't any spare time, hopefully it will be better in the weekend and I can answer all the above (to Günter anf Larry).
The machine I was running this on was a laptop with: Prosessor Intel(R) Core(TM) i7-10750H CPU @ 2.60GHz, 2592 Mhz, 6 kjerne(r), 12 logiske prosessor(er)
It have 16GB ram and I did run 6 matches in paralell all with single thread and ponder off.
For Dragon this isn't important since it wil play the same whatever your cpu or timecontrol is. I see I forgot to metion that it was 1+1 games. It is in the gamefile though.
For the other engines some (most) will play better with longer time and maybe some of the engines that was reported 'stalled' or 'lost on time' would be helped with longer timecontrol. I didn't include those engines in the list. I did use CuteChess (GUI) for all games.
I see now that I should probably used some sort of adjucating since Dragon on lower level have problem with winning with a queen up. In a human-engine game the human would have resigned long before and it is there the data hopefully could be used. All I was interested in from the other engines was that they did run without any error in 1+1 matches. To use adjucating you probably need to check the engines a little more to see that they have a reliable pv line.
I don't know how to run this bench you talk about, I could look at it in the weekend. All my use of Stockfish is through the web. SF 11 on Chess24 and SF 14 on Lichess. I see that Fritz (or chessbase) did come with Stockfish 13 so it is resting on my harddrive ready to be used.

Humans at very low level do not resign when they are queen down.
I believe that I probably can win most humans who know the chess rules with queen and 2 rooks odds.

The level of most humans who know to play chess is probably similiar to the level of the following player and I can probably beat this player and also slightly stronger players who can usually win against this player with queen and 2 rooks handicap.

https://www.chess.com/member/luigiguarino

I believe that most humans who know to play chess do not play in chess sites so the fact that this player is clearly weaker than big majority of the players who play in chess.com means nothing.

[Ferdy]

Running from PowerShell. First run after start. If I run it several time it can vary +/- 1-2%

Code: Select all

		
stockfish_10_x64_popcnt.exe
===========================
Total time (ms) : 1619
Nodes searched  : 3939338
Nodes/second    : 2433192

stockfish_10_x64_bmi2.exe
===========================
Total time (ms) : 1597
Nodes searched  : 3939338
Nodes/second    : 2466711

stockfish_10_x64.exe
===========================
Total time (ms) : 1725
Nodes searched  : 3939338
Nodes/second    : 2283674

stockfish_10_x32.exe
===========================
Total time (ms) : 2157
Nodes searched  : 3939338
Nodes/second    : 1826304

Edit: Changed it to fastest of 5-6 run.

All right your TC of 60+1 is not that far from CCRL Blitz 2+1. Now I can approximate the TC that I should use in my PC to get close to your result. Thanks.