size 6 tournament (unranked)

I just opened an unranked Havannah tournament for size 6.

(more info here)

Does "unranked" mean the same thing as "unrated"?

sorry, yes I meant "unrated".

What will be the hardwares and thinking times of the bots

in this tournament?

Ingo.

I see DF and RoF both participate ...

@Ingo: I don't put any restrictions.

It's not an official event by any means.

RoF will run on i7-920 (2.67 GHz, 12 GB)

and probably use around 5 min. per move

(the version currently playing uses 2 min.)

It can't use much more time than that

because then it'll soon run out of memory...

DeepFork will run also on i7. but for testing, only the single-core version is activated. The program generates 1 mio. UCT nodes (with 4 simulations for each new Node). So i think, on size 6, it will use 5-6 minutes per move.

I just registered and am going to play with an experimental version of Gambler with static evaluation. It is very slow, even on base-6 (just below 200 simulations per second) so in order to make it competitive it probably needs to find a factor 10-20 speedup.

Gambler will still be playing on the AMD Opteron 270 single cpu (never got around finalising the smp/cluster version) with 10 minutes per move.

Welcome Gambler! I'm looking forward to experiments and discussion on static evaluation (during or after the tournament). It seems Ring-of-Fire does need a lot of improvement there.

Does anyone know when this tournament actually starts? I put the date of today (September 17) but it seems it's not happening.

Don't worry about slow/expensive computation - it's the same with my program.

(On size 8, approx. 1000 playouts per second.)

my solution is to have a single-threaded main program,

and one thread per core doing playouts (independently, but all from

the same node at the frontier of the tree).

Since the playouts are heavy, this is quite OK.