CEGT - rating lists April 30th 2017

[Werner]

Hi all,

our actual rating lists are online and can be found under the attached links!

40 / 20:
New games: 2.240; 39 different engines
Total: 1.015.284

NEW Engines
584 Amoeba 2.4 x64: 2679 - 1000 games (+69 to v. 2.3)
332 Wasp 2.00 x64 1CPU: 2808 - 1000 games (+50 to v. 1.25)

UPDATES
2 Komodo 10.4 x64 4CPU: 3400 - 1593 games (-1)

40 / 120:
There was an update on 19th of April with 600 new games; now 26.120 games and 110 engines.
New results with Chiron 4.0 x64, Andscacs 0.89 x64, Fizbo 1.9 x64

5'+3'' pb=on
last update was on 19th of April with Booot 6.1 x64 (2948/2000 games). We are now testing Wasp 2.00 x64

40 / 4
no update; we are testing Gandalf 7.0, Amoeba 2.4 and Wasp 2.00
A big „Thank you“ to all testers as usual!!

Links

40/20: http://www.cegt.net/rating.htm
Blitz: http://www.cegt.net/blitz.htm
40/120: http://www.cegt.net/rating120.htm
Tester: http://www.cegt.net/testers/testers.htm
40/20 pb=on: http://www.cegt.net/rating4020PBON.htm
5+3 pb=on: http://www.cegt.net/rating5plus3pbon.htm
Games of the week: http://www.cegt.net/40_40%20Rating%20Li ... on/gow.jpg

Werner Schüle
CEGT-Team

[Frank Quisinsky]

Hi Werner,

thanks for Wasp testing.
I am looking, John too ...

I have 1.000 games with FCP Rating List conditions too at the moment. CEGT results with faster time controls should be better, think so, not sure.

I need here a lot of games for adjusting "Modul 1" of my new opening book. "Modul 1" are the basic moves with prioritys, "Modul 2" is the database engine have to analyzed I can copy later over the Modul 1 (import) and ready. So my current book for testing Wasp isn't perfect for testing but lesser as 1% bad lines are at the moment included.

Good news ...
Dirty is available ... I am very happy about it.

And nice to see that Gandalf 7 will be tested by CEGT. Maybe 2.600 Elo is possbile for the 10 years old Gandalf 7 engine with 1 core, hope so. OK, 2.625 are more real! Let see it ...

Best
Frank

[S.Taylor]

Hi all,

our actual rating lists are online and can be found under the attached links!

40 / 20:
New games: 2.240; 39 different engines
Total: 1.015.284

NEW Engines
584 Amoeba 2.4 x64: 2679 - 1000 games (+69 to v. 2.3)
332 Wasp 2.00 x64 1CPU: 2808 - 1000 games (+50 to v. 1.25)

UPDATES
2 Komodo 10.4 x64 4CPU: 3400 - 1593 games (-1)

40 / 120:
There was an update on 19th of April with 600 new games; now 26.120 games and 110 engines.
New results with Chiron 4.0 x64, Andscacs 0.89 x64, Fizbo 1.9 x64

5'+3'' pb=on
last update was on 19th of April with Booot 6.1 x64 (2948/2000 games). We are now testing Wasp 2.00 x64

40 / 4
no update; we are testing Gandalf 7.0, Amoeba 2.4 and Wasp 2.00
A big „Thank you“ to all testers as usual!!

Links

40/20: http://www.cegt.net/rating.htm
Blitz: http://www.cegt.net/blitz.htm
40/120: http://www.cegt.net/rating120.htm
Tester: http://www.cegt.net/testers/testers.htm
40/20 pb=on: http://www.cegt.net/rating4020PBON.htm
5+3 pb=on: http://www.cegt.net/rating5plus3pbon.htm
Games of the week: http://www.cegt.net/40_40%20Rating%20Li ... on/gow.jpg

Werner Schüle
CEGT-Team

I LIKE this!
Komodo 9.3, which i think is the version from the TCEC before last, is way above the Komodo version (10.2?) in the last TCEC.

So yet another support to my claim that version 10 was a downgrade instead of an upgrade.

I know the producers (Lefler, kaufman) vehemently deny this charge, but this list agrees with me.

Komodo 10 might have merits and potential, but perhaps it was not yet completed, and was still falling behind the older version.

[Uri Blass]

Hi all,

our actual rating lists are online and can be found under the attached links!

40 / 20:
New games: 2.240; 39 different engines
Total: 1.015.284

NEW Engines
584 Amoeba 2.4 x64: 2679 - 1000 games (+69 to v. 2.3)
332 Wasp 2.00 x64 1CPU: 2808 - 1000 games (+50 to v. 1.25)

UPDATES
2 Komodo 10.4 x64 4CPU: 3400 - 1593 games (-1)

40 / 120:
There was an update on 19th of April with 600 new games; now 26.120 games and 110 engines.
New results with Chiron 4.0 x64, Andscacs 0.89 x64, Fizbo 1.9 x64

5'+3'' pb=on
last update was on 19th of April with Booot 6.1 x64 (2948/2000 games). We are now testing Wasp 2.00 x64

40 / 4
no update; we are testing Gandalf 7.0, Amoeba 2.4 and Wasp 2.00
A big „Thank you“ to all testers as usual!!

Links

40/20: http://www.cegt.net/rating.htm
Blitz: http://www.cegt.net/blitz.htm
40/120: http://www.cegt.net/rating120.htm
Tester: http://www.cegt.net/testers/testers.htm
40/20 pb=on: http://www.cegt.net/rating4020PBON.htm
5+3 pb=on: http://www.cegt.net/rating5plus3pbon.htm
Games of the week: http://www.cegt.net/40_40%20Rating%20Li ... on/gow.jpg

Werner Schüle
CEGT-Team

I LIKE this!
Komodo 9.3, which i think is the version from the TCEC before last, is way above the Komodo version (10.2?) in the last TCEC.

So yet another support to my claim that version 10 was a downgrade instead of an upgrade.

I know the producers (Lefler, kaufman) vehemently deny this charge, but this list agrees with me.

Komodo 10 might have merits and potential, but perhaps it was not yet completed, and was still falling behind the older version.

last version is not 10.2 but it seems that you look only at the data that you like to loke

Komodo9.3 is above Komodo10.2
only at the 120/40 list that has almost no games and the following non significant results

3 Komodo 9.3 x64 3198 33 33 350 66.9% 3062 53.1%
5 Komodo 10.2 x64 3174 37 37 300 61.2% 3088 56.3%

If I look at the 20 minutes/40 moves list I see the following

2 Komodo 10.4 x64 4CPU 3400 16 16 1593 73.5% 3207
4 Komodo 10.2 x64 4CPU 3363 15 15 1785 70.9% 3193
12 Komodo 9.3 x64 4CPU 3324 16 16 1498 72.1% 3149


13 Komodo 10.4 x64 1CPU 3323 16 16 1300 69.4% 3163
20 Komodo 10.2 x64 1CPU 3268 15 15 1661 67.1% 3132
26 Komodo 9.3 x64 1CPU 3232 14 14 1740 70.7% 3064


If I look at the blitz list I see the following:

4 Komodo 10.4 x64 12CPU 3484 25 25 600 59.8% 3393
8 Komodo 10.2 x64 12CPU 3450 17 17 1400 64.3% 3330
14 Komodo 9.3 x64 12CPU 3414 18 18 1200 63.2% 3308


18 Komodo 10.4 x64 4CPU 3396 22 22 950 73.7% 3198
28 Komodo 9.3 x64 4CPU 3340 15 15 1400 67.7% 3201

37 Komodo 10.2 x64 2CPU 3311 27 27 690 77.8% 3053
41 Komodo 9.3 x64 2CPU 3305 19 19 1250 79.5% 3040


42 Komodo 10.4 x64 1CPU 3305 12 12 2400 70.3% 3141
52 Komodo 10.2 x64 1CPU 3273 13 13 1900 66.0% 3148
66 Komodo 9.3 x64 1CPU 3225 10 10 3100 69.5% 3070

[mjlef]

Hi all,

our actual rating lists are online and can be found under the attached links!

40 / 20:
New games: 2.240; 39 different engines
Total: 1.015.284

NEW Engines
584 Amoeba 2.4 x64: 2679 - 1000 games (+69 to v. 2.3)
332 Wasp 2.00 x64 1CPU: 2808 - 1000 games (+50 to v. 1.25)

UPDATES
2 Komodo 10.4 x64 4CPU: 3400 - 1593 games (-1)

40 / 120:
There was an update on 19th of April with 600 new games; now 26.120 games and 110 engines.
New results with Chiron 4.0 x64, Andscacs 0.89 x64, Fizbo 1.9 x64

5'+3'' pb=on
last update was on 19th of April with Booot 6.1 x64 (2948/2000 games). We are now testing Wasp 2.00 x64

40 / 4
no update; we are testing Gandalf 7.0, Amoeba 2.4 and Wasp 2.00
A big „Thank you“ to all testers as usual!!

Links

40/20: http://www.cegt.net/rating.htm
Blitz: http://www.cegt.net/blitz.htm
40/120: http://www.cegt.net/rating120.htm
Tester: http://www.cegt.net/testers/testers.htm
40/20 pb=on: http://www.cegt.net/rating4020PBON.htm
5+3 pb=on: http://www.cegt.net/rating5plus3pbon.htm
Games of the week: http://www.cegt.net/40_40%20Rating%20Li ... on/gow.jpg

Werner Schüle
CEGT-Team

I LIKE this!
Komodo 9.3, which i think is the version from the TCEC before last, is way above the Komodo version (10.2?) in the last TCEC.

So yet another support to my claim that version 10 was a downgrade instead of an upgrade.

I know the producers (Lefler, kaufman) vehemently deny this charge, but this list agrees with me.

Komodo 10 might have merits and potential, but perhaps it was not yet completed, and was still falling behind the older version.

last version is not 10.2 but it seems that you look only at the data that you like to loke

Komodo9.3 is above Komodo10.2
only at the 120/40 list that has almost no games and the following non significant results

3 Komodo 9.3 x64 3198 33 33 350 66.9% 3062 53.1%
5 Komodo 10.2 x64 3174 37 37 300 61.2% 3088 56.3%

If I look at the 20 minutes/40 moves list I see the following

2 Komodo 10.4 x64 4CPU 3400 16 16 1593 73.5% 3207
4 Komodo 10.2 x64 4CPU 3363 15 15 1785 70.9% 3193
12 Komodo 9.3 x64 4CPU 3324 16 16 1498 72.1% 3149


13 Komodo 10.4 x64 1CPU 3323 16 16 1300 69.4% 3163
20 Komodo 10.2 x64 1CPU 3268 15 15 1661 67.1% 3132
26 Komodo 9.3 x64 1CPU 3232 14 14 1740 70.7% 3064


If I look at the blitz list I see the following:

4 Komodo 10.4 x64 12CPU 3484 25 25 600 59.8% 3393
8 Komodo 10.2 x64 12CPU 3450 17 17 1400 64.3% 3330
14 Komodo 9.3 x64 12CPU 3414 18 18 1200 63.2% 3308


18 Komodo 10.4 x64 4CPU 3396 22 22 950 73.7% 3198
28 Komodo 9.3 x64 4CPU 3340 15 15 1400 67.7% 3201

37 Komodo 10.2 x64 2CPU 3311 27 27 690 77.8% 3053
41 Komodo 9.3 x64 2CPU 3305 19 19 1250 79.5% 3040


42 Komodo 10.4 x64 1CPU 3305 12 12 2400 70.3% 3141
52 Komodo 10.2 x64 1CPU 3273 13 13 1900 66.0% 3148
66 Komodo 9.3 x64 1CPU 3225 10 10 3100 69.5% 3070

That is exactly right. There are not enough games played to reach a conclusion at this time control for what Komodo version is strongest. Here is part of the table:

3 Komodo 9.3 x64 3198 33 33 350 66.9% 3062 53.1%
4 Stockfish 7.0 x64 3196 36 36 300 66.2% 3068 55.0%
5 Komodo 10.2 x64 3174 37 37 300 61.2% 3088 56.3%

the number 37 next to Komodo 10.2 are error margins. So in this case, within 2 standard deviations, Komodo could be 3174 +37 = 3211, or 3174-37 = 3134 elo. There is just not enough data to conclude whihc is stronger. But it is likely Komodo 10.2 is stronger than 9.3 based on the other rating lists. It takes a lot of time for the rating list guys to get enough games at these long time controls. They have to limit the versions they can test since there are just not enough machines available for the runs. If you happen to have a free cluster, let them know!

[Wolfgang]

Mark, you are completely right! Considering 40/120 I simply do not have enough hardware to test every single version or to let play thousands of games with those I test. I'm not happy with that, but what shall I do? A cluster would be fine but only together with a (nuclear) power plant...

So, S.Taylors conclusion that version 10 is weaker than version 9 is complete nonsense.

[Wolfgang]

....
So yet another support to my claim that version 10 was a downgrade instead of an upgrade.

I know the producers (Lefler, kaufman) vehemently deny this charge, but this list agrees with me.

Please stop using our list(s) as a so called "proof" for your conspiracy theories. Never read such a nonsense before.