[Pachi] pachi + dcnn

lemonsqueeze lemonsqueeze at free.fr
Tue Mar 29 15:37:54 CEST 2016


Hi,

So far i've tested with 180k simulations, 16 threads:

pachi -t =180000 
threads=16,resign_threshold=0.30,pass_all_alive,pondering=0 -f 
extra/opening.dat

With the opening book it's always playing the same opening but speeds up 
testing =)
No idea what the right dcnn_eqex might be for different settings, i've 
only tested with this so far. Even there it can still probably be 
improved (1400 looks good too, 1200 & 1500 definitely worse ...)

The github branch is a cleaned up version of the tree i'm using for 
testing, hopefully i didn't screw up (haven't tested it yet).
One thing that's missing in it is PR#26 (check group suicides), it can 
assert out in the endgame otherwise. Also right now dcnn disables 
pondering, would be nice to find a way to make it work again for 
tournaments.

I don't have a gpu so hard for me to experiment, but sure would be nice 
to have support for that ! Wow, it would really start to look like 
darkforest then =)
I'm thinking even multi-threaded cpu support might be good enough to 
start using it further down the tree...

Btw do you know what's going on with monkey jumps ?
For some reason it seems it never plays them, i suspect there's 
something strange going on... Playout issue maybe ?


On 03/29/2016 12:16 PM, Petr Baudis wrote:
>    HI,
>
>    that's awesome!  The eqex you said to give 88%, what time setting
> / threads# etc did you use for playtesting?  (It's usually scaling down
> when adding more computing power.  Just a FYI for people to run Pachi
> e.g. at tournaments.)
>
>    I'll be very happy to merge this, the implementation looks nicely
> clean!  It's just a matter of adding some README docs for this, I think.
>
>    If you find time to add async GPU support, I could run this on severs
> with 2x Tesla K20Xm e.g. for KGS tournaments, if you don't have a good
> machine for that.
>
> On Tue, Mar 29, 2016 at 11:04:12AM +0200, lemonsqueeze wrote:
>> Yes i still use patterns, and a large dcnn eqex so it gets a chance to
>> override it. Here's my branch on github:
>>
>>    https://github.com/lemonsqueeze/pachi/tree/dcnn
>>
>>
>>> Hi,
>>>
>>> I did exactly the same thing but not quite successful, about 60% win
>>> rate.
>>> Do you still use pattern? It will be great to have chance to see the
>>> details.
>>>
>>> Thanks.
>>
>>
>>
>> _______________________________________________
>> Pachi mailing list
>> Pachi at v.or.cz
>> http://rover.ms.mff.cuni.cz/mailman/listinfo/pachi
>>
>


More information about the Pachi mailing list