Quo Vadis Houdini, Stockfish, Komodo

[Hugo]

hi ALL

actually I am playing a match between Houdini 6.03 pext, Stockfish 180218 Bmi2 and Komodo 2016.00 Bmi2.
Time for a game is 45 min + 10sec., 75rounds/225 games

Engines use 15 cpu, full set of Syzygy tablebases, hashtables 8192MB
ponder is ON
Shredder GUI, small 5mvs book
Games are running on a E5-2697A v4 machine (two 16 core cpu on board).

Fist games can be reviewed here
http://www.clemens-keck.de/ReplayZone/quovadis.htm

crosstable
http://www.clemens-keck.de/ReplayZone/QuoVadis.html

live view, update frequenzy 30 sec
http://www.clemens-keck.de/screen/


regards, Clemens Keck

[Hugo]

Hi All

tournament is still running.
Standings after 122 games:

Code: Select all

1	Stockfish 180218 64 BMI2 t15     43.5 / 82
2	Komodo 2016.00 64-bit t15        41.0 / 81
3	Houdini 6.03 Pro x64-pext t15    37.5 / 81

since round 39.2 I use the feobos c3 book. Am curious about the games.
All files updated and live view too.

regards, C.K.

[Carlos777]

Thanks for this interesting tournament.

Contempt is default for everyone?

[Hugo]

Hi,

Komodo contempt is 0.
All others contempt is default.

regards, C.K.

[Jouni]

In this and YLCET tournament Komodo is now beating Houdini. But TCEC is different so far with older 2012.00 version.

[Nordlandia]

Clemens Keck: Mark Lefler said in tcec chat that PB=enabled on dual xeon system is not necessarily worse off than PB=off. What is your impression with ponder enabled so far?

[Hugo]

ponder off is very entertaining
You can see and compare the engines calculation.
15 cores is really enough for a good performance, and ponder on tournaments are seldom.
Komodo does well so far, seems to be improved.

C.K.

[Nordlandia]

From Houdini manual:

The objectively highest level of play is achieved without contempt, i.e. Contempt value 0.

A contempt value of 2 will typically reduce the draw rate in a match by about 4%.

So for tournament matches against K and SF, C=0 is better choise against default.

H with C=0 might perform better in same ongoing match i suppose.

[Hugo]

Thank you ! This is quite intresting, and I will consider this new tournaments.

C.K.

From Houdini manual:

The objectively highest level of play is achieved without contempt, i.e. Contempt value 0.

A contempt value of 2 will typically reduce the draw rate in a match by about 4%.

So for tournament matches against K and SF, C=0 is better choise against default.

H with C=0 might perform better in same ongoing match i suppose.