World Championship Winning Computer Chess Software Program & Downloads for Chess Databases, Analysis and Play on PC, Mac, iPad and iPhone — Visit: Hiarcs.com
gordonr wrote: ↑Wed Apr 24, 2024 10:54 amThe test games can be converted to a list of EPDs. Then within MessChess/Arena, it's possible to do auto analysis of the list of EPDs.
Hi Gordon,
yes, you are right. It may not be as fun as playing the moves by yourself, but indeed it is a way to get the list of played moves thru some automation.
It is just a GUI feature.
gordonr wrote: ↑Wed Apr 24, 2024 10:54 am
It's then possible to write a short piece of code that will parse this output and score the moves, etc along with the other calculations.
Parsing the output to create an easier to use list of moves, aiming at more efficiently feeding the spreadsheets can as well be done.
But scoring the moves and running other calculations within such a piece of code would require forking the master piece of work that Nick developped, and coded undisclosed within the Excel spreadsheets. Let's keep the fun running the test "as is"...
Warm regards,
Eric
In some of the game records tabs I had a few that we locked too strongly which would stop you from using those tabs as game records. Sorry about that. Please go back and reupload the test sheets.
gordonr wrote: ↑Wed Apr 24, 2024 4:28 pm
I tried testing game 1 with Mephisto Atlanta at 30 seconds/move. That was interesting to go through.
Hi Gordon,
See as I wrote about earlier the tests should be easy, fast and fun, well not so fast at 3 minutes. Atlanta's game 1 score is in line with all the other tested ones as game 1 scores highly for most of them. You should have fun seeing Atlanta struggle through some games and get excited in others when it scores well.
Keep posting and I will collect and add to the master list that I will share periodically with all completed tests.
PS. perhaps you can paste them as a text list so I can grab them. A jpg won't let me do that.
Btw, Atlanta's weaker brother GK2000 scored higher than expected, therefore being a Morsch I suspect Atlanta will as well. In a perfect world it should come behind TM Lyon and even V11, but I suspect since V11 scored badly in one of the games and needs more games to catch up that Atlanta will score above V11. It will be interesting to see if it also outscores TM Lyon.
Sorry I must have made a mistake with the 30 sec test. 19. ... Bxf4 is not reproducible. 19. ... Qe4+ is the move at 30 secs. Are you ok to change that?
Here are the results for 3 mins. For further results, I'll batch them up more - multiple game results per post.
Mephisto
Atlanta
Frans Morsch
3 minutes
SH7034 - 32 bit - 20Mhz - 64 KB ROM - 512 KB RAM
TOTAL SCORE:
2693
WHITE MOVES
5.Nf3
6.Bxf7+
7.Qxf3
8.Qxb7
9.Nb5
10.d4
11.Bb5+
12.Qc6+
13.Qc6+
14.Qxb5+
15.0-0
16.Qxd5
17.Qxd5
18.c3
19.0-0
20.Qxd7+
WHITE SCORE:
2912
BLACK MOVES
4. ... e6
5. ... e6
6. ... Bh5
7. ... c6
8. ... Nbd7
9. ... Bd6
10. ... Rb8
11. ... Nxc4
12. ... Be7
13. ... Ke7
14. ... Nd7
15. ... exd5
16. ... c5
17. ... 0-0
18. ... c6
19. ... Qe4+
BLACK SCORE:
2473
TOTAL SCORE:
2693
Sorry I must have made a mistake with the 30 sec test. 19. ... Bxf4 is not reproducible. 19. ... Qe4+ is the move at 30 secs. Are you ok to change that?
Hi Gordon,
No problems, I corrected it. Atlanta score at 30S is now 2735.
I reuploaded the set so can you download it again. Atlanta is added and I made a correction in Game 9 where the FEN was incorrect.
So It might be best if you discard your old sheets and use what I reuploaded to be UpToDate with everything. The download Link remains the same in page 1 of these posts.
Eric found a couple dropdown glitches on game 6 and game 10. These are now fixed. Can you all please go back to page 1 and download the zips again and use the latest sheets. Hopefully we are now completely bug free. If you do spot one, then please let me know.
As 5 of the tests go back to 2017 but revamped with new evaluations, I spotted that Atlanta was played at 30 seconds by someone back then, It could have been me or someone else as several people did some tests.
Interesting is that there are some differences so it's possible that Atlanta's move randomness kick's in on occasions.
GAME 1
Gordon Old Test
MEPHISTO MEPHISTO
ATLANTA ATLANTA
FRANS MORSCH FRANS MORSCH
30 SECONDS 30 SECONDS
SH7034 - 32 bit - 20Mhz - 64 KB ROM - 512 KB RAM SH7034 - 32 bit - 20Mhz - 64 KB ROM - 512 KB RAM
TOTAL SCORE: TOTAL SCORE:
2735 2672
WHITE MOVES WHITE MOVES
5.Nf3 5.Nf3
6.Bxf7+ 6.Bxf7+
7.Qxf3 7.Qxf3
8.Qxb7 8.Qxb7
9.Nb5 9.Nb5
10.Nxa7 10.Nxa7
11.Bb5+ 11.Bb5+
12.Qc6+ 12.Bb5+
13.Qc6+ 13.Qc6+
14.Qxb5+ 14.Qxb5+
15.0-0 15.0-0
16.Qxd5 16.Qxd5
17.Qxd5 17.Qxd5
18.c3 18.c3
19.0-0 19.0-0
20.Qxd7+ 20.Qxd7+
WHITE SCORE: WHITE SCORE:
3224 3256
BLACK MOVES BLACK MOVES
4. ... e6 4. ... Nf6
5. ... Bg4 5. ... e6
6. ... Bh5 6. ... Bh5
7. ... c6 7. ... c6
8. ... Nbd7 8. ... Nbd7
9. ... Bd6 9. ... Bd6
10. ... Rb8 10. ... Rb8
11. ... Nxc4 11. ... Nxc4
12. ... Nd6 12. ... Nd6
13. ... Ke7 13. ... Ke7
14. ... Nd7 14. ... Nd7
15. ... exd5 15. ... exd5
16. ... c5 16. ... c5
17. ... 0-0 17. ... 0-0
18. ... c6 18. ... c6
19. ... Qe4+ 19. ... Qe6+
BLACK SCORE: BLACK SCORE:
2245 2089
TOTAL SCORE: TOTAL SCORE:
2735 2672
spacious_mind wrote: ↑Thu Apr 25, 2024 9:39 pm
Interesting is that there are some differences so it's possible that Atlanta's move randomness kick's in on occasions.
Interesting indeed. The randomness is off by default so I think that would have been the same. Pondering is on by default - do we know if anything can be updated in the hash tables while Atlanta is pondering?!
Two moves jump out at me as not looking right: 5. ... Bg4 and 19. ... Qe6+. Both of these are big and relatively simple mistakes that I believe Atlanta should always avoid at 30s. Maybe we can both check how reproducible they are. The 5. ... Bg4 could have been an input mistake by me since it's the game move. I done another 2 games today and I'm now in more of a habit of double checking.
spacious_mind wrote: ↑Thu Apr 25, 2024 9:39 pm
Interesting is that there are some differences so it's possible that Atlanta's move randomness kick's in on occasions.
Interesting indeed. The randomness is off by default so I think that would have been the same. Pondering is on by default - do we know if anything can be updated in the hash tables while Atlanta is pondering?!
Two moves jump out at me as not looking right: 5. ... Bg4 and 19. ... Qe6+. Both of these are big and relatively simple mistakes that I believe Atlanta should always avoid at 30s. Maybe we can both check how reproducible they are. The 5. ... Bg4 could have been an input mistake by me since it's the game move. I done another 2 games today and I'm now in more of a habit of double checking.
Yes, I think I am going to go with your results, those moves do look weird.
Sometimes it's worth a couple of people doing the tests as errors could be spotted.
moves 4 and 5 score the same for black the move that matters is 19. ... Qe6+ and I am pretty sure that was an error and should have been Qe4+
here are additional scores: Fidelity Chess Challenger 7 and Novag Chess Champion Super System III, at tournament level (bottom of the snapshot).
Looks fine, to me.
From active to tournament level, CC7 gained 112 points; CCSSIII 93.
BR,
Eric
Attachments
Renaissance_Tournament.jpg (171.52 KiB) Viewed 161 times