6cpu engine rapid marathon

[Hugo]

Hi all

I started a tournament with 13 engines. I am intrested to see what happens on average hardware setup.
conditions:
engines 6 cpu, no HT no learning, 4GB hashtables
ponder OFF
time control 20 min + 5 sec
no tablebases
Harry Schnapp book for shredder GUI
20 rounds = 1560 games to play
hardware 2 x Intel E5-2697A v4 (32cpu)
5 games at the same time, 2 cpu left for service and ftp trigger.

Participants:
Andscacs 0.88b r1 x6
Booot 6.1_x64_popcnt x6
Deep Shredder 161006 x64 x6
Fire 5 x64 x6
Fizbo 1.8 x6
Ginkgo 1.9m x6
Hannibal 1.7 x64 x6
Houdini 5.01 x64-pext x6
Jonny 8.00 x6
Komodo 10.2 64-bit c0 x6
Nirvanachess 2.3 x6
Stockfish 8 64 BMI2 x6
Texel 1.06 x6

crosstable upload frequency 10 minutes

http://clemens-keck.de/livegames/X6-tour.html

regards, Clemens Keck

[Vinvin]

Thanks, Clemens !
I'm following the race

[Frank Quisinsky]

Hi Clemens,

that is really super interesting.
For myself the most interesting Event this year for eng-eng I can follow by others.

Sure here that Komodo will win!
Have many thanks!

Best
Frank

Only a bit boring that not available engines can play here. Places for available engines are more interesting ... but you know my opinion about it. Not important ... fact is ... the tournament is more as interesting only.

[Jouni]

Houdini leads. Houdini seems to play relatively better with more cores than single CPU also in my tests! Better SMP implementation? But 44 TCEC cores is already too much?

[Ozymandias]

Has Houdini scored better against the bottom 5?

[JJJ]

Houdini won the TCEC rapid event too. So maybe it is better than Stockfish against more weak opponent on long time control and some cores.

[Lyudmil Tsvetkov]

Houdini won the TCEC rapid event too. So maybe it is better than Stockfish against more weak opponent on long time control and some cores.

more likely reason is the short opening book, where deeper search decreases in importance and eval discretion takes precedence.

the book used by Clemens is very good indeed, one of the best short books I have seen around: not many simplifications, not many important decisions taken on behalf of the engine by the book, very nice variety. (the TCEC book, for example, although interesting, simplifies too much, often takes important decisions on behalf of the engines, and has too many Sicilians)

[JJJ]

Yes it could be that.

[Hugo]

congratulations to Houdini 5 for winning the 6cpu marathon



i am already runninging a new tour with 12 cpu and 40 /120 + rest 60
will post soon.

regards, Clemens keck
http://clemens-keck.de/livegames/X6-tour.html

[cdani]

congratulations to Houdini 5 for winning the 6cpu marathon

i am already runninging a new tour with 12 cpu and 40 /120 + rest 60
will post soon.

regards, Clemens keck
http://clemens-keck.de/livegames/X6-tour.html

Thanks for the tournament!! Those tournaments with more cpus and time are really interesting.
My current Andscacs developing version seems clearly stronger (proved against other engines). Maybe you want to use it. Anything just tell me.