My test for Orion 0.7 NNUE_avx2_bmi2

Discussion of computer chess matches and engine tournaments.

Moderators: hgm, Rebel, chrisw

User avatar
Sylwy
Posts: 4468
Joined: Fri Apr 21, 2006 4:19 pm
Location: IASI - the historical capital of MOLDOVA
Full name: SilvianR

My test for Orion 0.7 NNUE_avx2_bmi2

Post by Sylwy »

Test conditions:

-TC=4'+2"
-Hash=256 MB
-GUI: Arena 3.5.1
-Book: IM_4mvs.abk (8 plies book) for Orion 0.7 NNUE_avx2_bmi2 and cheng2015.cb for Cheng 4.39_x64
-default settings for both engines
-1 thread-CPU=Intel i5-7400-3GHz (Kaby Lake)
-6-men Syzygy bases for both engines.

Image

+252 Elo points for Orion 0.7 NNUE_avx2_bmi2. Cheng 4.39_x64 has 2943 Elo (CCRL Blitz rating list).

Next: Koivisto 2.0_x64_bmi2 vs. Gogobello 2.2_x64_bmi2:

Image
mar
Posts: 2559
Joined: Fri Nov 26, 2010 2:00 pm
Location: Czech Republic
Full name: Martin Sedlak

Re: My test for Orion 0.7 NNUE_avx2_bmi2

Post by mar »

that's more like +400 elo for Orion, since it's listed as 2738 in CCRL blitz
Martin Sedlak
User avatar
Sylwy
Posts: 4468
Joined: Fri Apr 21, 2006 4:19 pm
Location: IASI - the historical capital of MOLDOVA
Full name: SilvianR

Re: My test for Orion 0.7 NNUE_avx2_bmi2

Post by Sylwy »

mar wrote: Tue Sep 22, 2020 4:44 pm that's more like +400 elo for Orion, since it's listed as 2738 in CCRL blitz
Yes ! Seems that CCRL is not far from my result:

Image

The same for RubiChess 1.9-NNUE. Andreas Matthies was right. The engine is just a player. The neural network makes the music.
User avatar
Guenther
Posts: 4607
Joined: Wed Oct 01, 2008 6:33 am
Location: Regensburg, Germany
Full name: Guenther Simon

Re: My test for Orion 0.7 NNUE_avx2_bmi2

Post by Guenther »

mar wrote: Tue Sep 22, 2020 4:44 pm that's more like +400 elo for Orion, since it's listed as 2738 in CCRL blitz
I will never understand the lazyness(? - actually a pic needs more work) or sloppiness to just show an unrecognizable
picture instead of a clear formatted table. Moreover in the future the pic will be gone into nirvana, while the
ascii table will remain as long as the forum exists.
In the pic above no one even knows, if Cheng4 or Orion has won the match...
https://rwbc-chess.de

trollwatch:
Talkchess nowadays is a joke - it is full of trolls/idiots/people stuck in the pleistocene > 80% of the posts fall into this category...
Terje
Posts: 347
Joined: Tue Nov 19, 2019 4:34 am
Location: https://github.com/TerjeKir/weiss
Full name: Terje Kirstihagen

Re: My test for Orion 0.7 NNUE_avx2_bmi2

Post by Terje »

Sylwy wrote: Tue Sep 22, 2020 5:54 pm
mar wrote: Tue Sep 22, 2020 4:44 pm that's more like +400 elo for Orion, since it's listed as 2738 in CCRL blitz
Yes ! Seems that CCRL is not far from my result:

Image

The same for RubiChess 1.9-NNUE. Andreas Matthies was right. The engine is just a player. The neural network makes the music.
Would be cool if someone ported the entire SF classical evaluation to their engine to see how much elo that gains. Would it be that different from using a SF net?
User avatar
Sylwy
Posts: 4468
Joined: Fri Apr 21, 2006 4:19 pm
Location: IASI - the historical capital of MOLDOVA
Full name: SilvianR

Re: My test for Orion 0.7 NNUE_avx2_bmi2

Post by Sylwy »

Guenther wrote: Tue Sep 22, 2020 5:56 pm
mar wrote: Tue Sep 22, 2020 4:44 pm that's more like +400 elo for Orion, since it's listed as 2738 in CCRL blitz
1.-I will never understand the lazyness(? - actually a pic needs more work) or sloppiness to just show an unrecognizable
picture instead of a clear formatted table.
2.-Moreover in the future the pic will be gone into nirvana, while the ascii table will remain as long as the forum exists.
3.-In the pic above no one even knows, if Cheng4 or Orion has won the match...
1.-Sorry ! That's my style - I like the colors, the logos, the pics.......nothing to do ! :lol:
2.-WOW ! Nirvana ? Good ! So......ascii table will remain on Earth's hell ? Nice ! :lol:
3.-Just a click and every pic will be enlarged. Not so hard !
User avatar
Guenther
Posts: 4607
Joined: Wed Oct 01, 2008 6:33 am
Location: Regensburg, Germany
Full name: Guenther Simon

Re: My test for Orion 0.7 NNUE_avx2_bmi2

Post by Guenther »

Sylwy wrote: Tue Sep 22, 2020 6:11 pm
Guenther wrote: Tue Sep 22, 2020 5:56 pm
mar wrote: Tue Sep 22, 2020 4:44 pm that's more like +400 elo for Orion, since it's listed as 2738 in CCRL blitz
1.-I will never understand the lazyness(? - actually a pic needs more work) or sloppiness to just show an unrecognizable
picture instead of a clear formatted table.
2.-Moreover in the future the pic will be gone into nirvana, while the ascii table will remain as long as the forum exists.
3.-In the pic above no one even knows, if Cheng4 or Orion has won the match...
1.-Sorry ! That's my style - I like the colors, the logos, the pics.......nothing to do ! :lol:
2.-WOW ! Nirvana ? Good ! So......ascii table will remain on Earth's hell ? Nice ! :lol:
3.-Just a click and every pic will be enlarged. Not so hard !
Just in case you still do not realize:

The players names in the result details are cut off.
The table header says Orion vs. Cheng 19.5 : 80.5.
Enlarging doesn't help here.
Is that so hard to see?

Your lols are out of place.
https://rwbc-chess.de

trollwatch:
Talkchess nowadays is a joke - it is full of trolls/idiots/people stuck in the pleistocene > 80% of the posts fall into this category...
User avatar
Sylwy
Posts: 4468
Joined: Fri Apr 21, 2006 4:19 pm
Location: IASI - the historical capital of MOLDOVA
Full name: SilvianR

Re: My test for Orion 0.7 NNUE_avx2_bmi2

Post by Sylwy »

Guenther wrote: Tue Sep 22, 2020 7:07 pm The players names in the result details are cut off.
The table header says Orion vs. Cheng 19.5 : 80.5.
Enlarging doesn't help here.
Good ? A new pic.

Image

Was the single complaint (on this subject, of course :) ) in 17 years (2006-2020 & 2003-2006-the old talkchess forum) !
User avatar
Guenther
Posts: 4607
Joined: Wed Oct 01, 2008 6:33 am
Location: Regensburg, Germany
Full name: Guenther Simon

Re: My test for Orion 0.7 NNUE_avx2_bmi2

Post by Guenther »

Sylwy wrote: Tue Sep 22, 2020 7:28 pm
Guenther wrote: Tue Sep 22, 2020 7:07 pm The players names in the result details are cut off.
The table header says Orion vs. Cheng 19.5 : 80.5.
Enlarging doesn't help here.
Good ? A new pic.

Image

Was the single complaint (on this subject, of course :) ) in 17 years (2006-2020 & 2003-2006-the old talkchess forum) !
Much better, except the low half-time of course ;-)

this is how such things look in 6 years
http://talkchess.com/forum3/viewtopic.p ... 92#p561592

from three years ago
http://talkchess.com/forum3/viewtopic.p ... 89&t=50060

anyhow greets from Regensburg :)
https://rwbc-chess.de

trollwatch:
Talkchess nowadays is a joke - it is full of trolls/idiots/people stuck in the pleistocene > 80% of the posts fall into this category...
User avatar
Sylwy
Posts: 4468
Joined: Fri Apr 21, 2006 4:19 pm
Location: IASI - the historical capital of MOLDOVA
Full name: SilvianR

Re: My test for Orion 0.7 NNUE_avx2_bmi2

Post by Sylwy »

Guenther wrote: Tue Sep 22, 2020 7:43 pm
Sylwy wrote: Tue Sep 22, 2020 7:28 pm
Guenther wrote: Tue Sep 22, 2020 7:07 pm The players names in the result details are cut off.
The table header says Orion vs. Cheng 19.5 : 80.5.
Enlarging doesn't help here.
Good ? A new pic.

Image

Was the single complaint (on this subject, of course :) ) in 17 years (2006-2020 & 2003-2006-the old talkchess forum) !
Much better, except the low half-time of course ;-)

this is how such things look in 6 years
http://talkchess.com/forum3/viewtopic.p ... 92#p561592

from three years ago
http://talkchess.com/forum3/viewtopic.p ... 89&t=50060

anyhow greets from Regensburg :)
Nothing is eternal in this world (even the websites hosting services..........even a lot of free chess engines which no one finds on the web today....me, for example, I'm very interested to have an old chess engine : Octavius by Luke Pellen-an engine having an evaluation based on a four-layer neural network-1999 year........was free in epoch....nobody have it today-that's the true drama, not my tests).

Greets from IASI and a nice gift-a video from my YouTube channel :) :