Torch V1 - Chess324 Testing Results

Discussion of computer chess matches and engine tournaments.

Moderator: Ras

Modern Times
Posts: 3699
Joined: Thu Jun 07, 2012 11:02 pm

Torch V1 - Chess324 Testing Results

Post by Modern Times »

https://ccrl.chessdom.com/ccrl/Chess324/index.html

Just to recap, this is a Top 15 revolving round-robin tournament using SPCC's Anti-Draw Chess324 opening book - SPCC 3-move Chess324_xxl_big_+090_+119. The book contains 70355 lines. Each pairing plays 100 openings reversed sides, so 200 games per pairing. Thanks to Stefan for his excellent work in creating these Chess324 books.

Revolving means that when a new engine version comes out, the old one is removed and the new one added. In this case also, because Torch came in as a new Top 15 engine, Seer at the bottom fell out.

This tournament was first published back in August.
viewtopic.php?f=6&t=82500


Torch is the #2 engine under these test conditions.
Top
Modern Times
Posts: 3699
Joined: Thu Jun 07, 2012 11:02 pm

Re: Torch V1 - Chess324 Testing Results

Post by Modern Times »

Code: Select all

Rank        Engine         Elo   +    -   Score  AvOp  Games
 1 Stockfish 20230813   3787  +10  -11  74.9% -161.8  2800
 2 Torch v1             3761  +10  -10  71.0% -133.9  2800
 3 Dragon by Komodo 3.3 3735  +10  -10  67.0% -106.1  2800
 4 Berserk 11.1         3673  +10  -10  56.7%  -39.6  2800
 5 Ethereal 14.00 FRC   3656  +10  -10  53.6%  -21.4  2800
 6 RubiChess 20230918   3632  +10  -10  49.3%   +4.3  2800
 7 Caissa 1.13.1        3623  +10  -10  47.9%  +13.9  2800
 8 CS Tal 2.00 Elo      3620  +10  -10  47.5%  +17.1  2800
 9 Clover 6.0           3612  +10  -10  45.9%  +25.7  2800
10 Koivisto 9.0         3598  +10  -10  43.6%  +40.7  2800
11 Rebel EAS            3580  +10  -10  40.4%  +60.0  2800
12 rofChade 3.1         3579  +10  -10  40.3%  +61.1  2800
13 Igel 3.5.0           3564  +10  -10  37.7%  +77.1  2800
14 Revenge 3.0          3563  +10  -10  37.6%  +78.2  2800
15 Viridithas 11.0.0    3557  +10  -10  36.6%  +84.6  2800
Top
Modern Times
Posts: 3699
Joined: Thu Jun 07, 2012 11:02 pm

Re: Torch V1 - Chess324 Testing Results

Post by Modern Times »

If you prefer Ordo, using CS Tal 2.00 Elo 3620 as the point of equivalence in both:

Code: Select all

# PLAYER                  :  RATING  POINTS  PLAYED   (%)
   1 Stockfish 20230813      :  3824.2  2096.0    2800    75
   2 Torch v1                :  3791.3  1987.5    2800    71
   3 Dragon by Komodo 3.3    :  3759.7  1876.0    2800    67
   4 Berserk 11.1            :  3684.4  1588.0    2800    57
   5 Ethereal 14.00 FRC      :  3662.5  1500.5    2800    54
   6 RubiChess 20230918      :  3633.0  1381.5    2800    49
   7 Caissa 1.13.1           :  3623.1  1341.5    2800    48
   8 CS Tal 2.00 Elo         :  3620.0  1329.0    2800    47
   9 Clover 6.0              :  3609.1  1285.0    2800    46
  10 Koivisto 9.0            :  3593.4  1222.0    2800    44
  11 Rebel EAS               :  3570.7  1132.0    2800    40
  12 rofChade 3.1            :  3569.9  1128.5    2800    40
  13 Igel 3.5.0              :  3551.1  1055.5    2800    38
  14 Revenge 3.0             :  3550.1  1051.5    2800    38
  15 Viridithas 11.0.0       :  3543.3  1025.5    2800    37
Top
Chessqueen
Posts: 5685
Joined: Wed Sep 05, 2018 2:16 am
Location: Moving
Full name: Jorge Picado

Re: Torch V1 - Chess324 Testing Results

Post by Chessqueen »

Modern Times wrote: ↑Sat Oct 14, 2023 10:23 pm If you prefer Ordo, using CS Tal 2.00 Elo 3620 as the point of equivalence in both:

Code: Select all

# PLAYER                  :  RATING  POINTS  PLAYED   (%)
   1 Stockfish 20230813      :  3824.2  2096.0    2800    75
   2 Torch v1                :  3791.3  1987.5    2800    71
   3 Dragon by Komodo 3.3    :  3759.7  1876.0    2800    67
   4 Berserk 11.1            :  3684.4  1588.0    2800    57
   5 Ethereal 14.00 FRC      :  3662.5  1500.5    2800    54
   6 RubiChess 20230918      :  3633.0  1381.5    2800    49
   7 Caissa 1.13.1           :  3623.1  1341.5    2800    48
   8 CS Tal 2.00 Elo         :  3620.0  1329.0    2800    47
   9 Clover 6.0              :  3609.1  1285.0    2800    46
  10 Koivisto 9.0            :  3593.4  1222.0    2800    44
  11 Rebel EAS               :  3570.7  1132.0    2800    40
  12 rofChade 3.1            :  3569.9  1128.5    2800    40
  13 Igel 3.5.0              :  3551.1  1055.5    2800    38
  14 Revenge 3.0             :  3550.1  1051.5    2800    38
  15 Viridithas 11.0.0       :  3543.3  1025.5    2800    37
So, Torch is rated between Stockfish and Dragon 3.3 ? And how long it will be before it is stronger than Stockfish ?
Top
CornfedForever
Posts: 648
Joined: Mon Jun 20, 2022 4:08 am
Full name: Brian D. Smith

Re: Torch V1 - Chess324 Testing Results

Post by CornfedForever »

Chessqueen wrote: ↑Sun Oct 15, 2023 3:18 am
Modern Times wrote: ↑Sat Oct 14, 2023 10:23 pm If you prefer Ordo, using CS Tal 2.00 Elo 3620 as the point of equivalence in both:

Code: Select all

# PLAYER                  :  RATING  POINTS  PLAYED   (%)
   1 Stockfish 20230813      :  3824.2  2096.0    2800    75
   2 Torch v1                :  3791.3  1987.5    2800    71
   3 Dragon by Komodo 3.3    :  3759.7  1876.0    2800    67
   4 Berserk 11.1            :  3684.4  1588.0    2800    57
   5 Ethereal 14.00 FRC      :  3662.5  1500.5    2800    54
   6 RubiChess 20230918      :  3633.0  1381.5    2800    49
   7 Caissa 1.13.1           :  3623.1  1341.5    2800    48
   8 CS Tal 2.00 Elo         :  3620.0  1329.0    2800    47
   9 Clover 6.0              :  3609.1  1285.0    2800    46
  10 Koivisto 9.0            :  3593.4  1222.0    2800    44
  11 Rebel EAS               :  3570.7  1132.0    2800    40
  12 rofChade 3.1            :  3569.9  1128.5    2800    40
  13 Igel 3.5.0              :  3551.1  1055.5    2800    38
  14 Revenge 3.0             :  3550.1  1051.5    2800    38
  15 Viridithas 11.0.0       :  3543.3  1025.5    2800    37
So, Torch is rated between Stockfish and Dragon 3.3 ? And how long it will be before it is stronger than Stockfish ?
I suggest you need to go to a carnival and consult a soothsayer.
Top
User avatar
Rebel
Posts: 7297
Joined: Thu Aug 18, 2011 12:04 pm
Full name: Ed Schröder

Re: Torch V1 - Chess324 Testing Results

Post by Rebel »

Modern Times wrote: ↑Sat Oct 14, 2023 10:23 pm If you prefer Ordo, using CS Tal 2.00 Elo 3620 as the point of equivalence in both:

Code: Select all

# PLAYER                  :  RATING  POINTS  PLAYED   (%)
   1 Stockfish 20230813      :  3824.2  2096.0    2800    75
   2 Torch v1                :  3791.3  1987.5    2800    71
   3 Dragon by Komodo 3.3    :  3759.7  1876.0    2800    67
   4 Berserk 11.1            :  3684.4  1588.0    2800    57
   5 Ethereal 14.00 FRC      :  3662.5  1500.5    2800    54
   6 RubiChess 20230918      :  3633.0  1381.5    2800    49
   7 Caissa 1.13.1           :  3623.1  1341.5    2800    48
   8 CS Tal 2.00 Elo         :  3620.0  1329.0    2800    47
   9 Clover 6.0              :  3609.1  1285.0    2800    46
  10 Koivisto 9.0            :  3593.4  1222.0    2800    44
  11 Rebel EAS               :  3570.7  1132.0    2800    40
  12 rofChade 3.1            :  3569.9  1128.5    2800    40
  13 Igel 3.5.0              :  3551.1  1055.5    2800    38
  14 Revenge 3.0             :  3550.1  1051.5    2800    38
  15 Viridithas 11.0.0       :  3543.3  1025.5    2800    37
In a robin-round you don't need anchors.

For the ADRL I am just using : -a 3500
90% of coding is debugging, the other 10% is writing bugs.
Top
Modern Times
Posts: 3699
Joined: Thu Jun 07, 2012 11:02 pm

Re: Torch V1 - Chess324 Testing Results

Post by Modern Times »

Rebel wrote: ↑Sun Oct 15, 2023 7:59 am
In a robin-round you don't need anchors.

For the ADRL I am just using : -a 3500
I meant, on the bayeselo list CS Tal 2.00 Elo was 3620, so I made that engine also 3620 on the Ordo list so the two were comparable.
Top
User avatar
Rebel
Posts: 7297
Joined: Thu Aug 18, 2011 12:04 pm
Full name: Ed Schröder

Re: Torch V1 - Chess324 Testing Results

Post by Rebel »

I see... but it's of course not the real elo of CSTAL in this case.
90% of coding is debugging, the other 10% is writing bugs.
Top
Frank Quisinsky
Posts: 6888
Joined: Wed Nov 18, 2009 7:16 pm
Location: Gutweiler, Germany
Full name: Frank Quisinsky

Re: Torch V1 - Chess324 Testing Results

Post by Frank Quisinsky »

Hi Ed,

maybe CSTal need more time for a higher Elo.
Same for Igel !!

If I am looking on CCRL 40/15 Igel and CSTal have much better rankings!

Same for your engine!
In CCRL 40/15 a much better ranking!

I believe in my tourney Rebel EAS will go higher as the current standing, because I am using with 40 in 20 and 4Ghz a higher time-control. At the moment I wonder a bit about the ranking in my tourney. Ranking 6 or 7 should be possible.

The Torch results are really fantastic.
Congratulations!

I remember ...
Houdini are a long time better as Stockfish and Komodo.

Best
Frank
Top
Modern Times
Posts: 3699
Joined: Thu Jun 07, 2012 11:02 pm

Re: Torch V1 - Chess324 Testing Results

Post by Modern Times »

Rebel wrote: ↑Sun Oct 15, 2023 10:59 am I see... but it's of course not the real elo of CSTAL in this case.
Well, that is the question - there is no "real Elo" in computer ratings lists, the ratings list publishers just settle on an "offset" or "anchor" or whatever you want to call it. CS Tal may be 3500 on one list, 3600 on another, 3700 on yet another etc.

In this Chess324 tourney, I set the Elo values to match the standard chess blitz list for that basket of engines. So on the CCRL standard blitz list, CS Tal is 3618 Elo

http://ccrl.chessdom.com/ccrl/404/cgi/c ... librate=no
Top

Return to “Computer Chess Club: Tournaments and Matches”