I've updated my test script to include those statistics.  Oddly enough,
it ran significantly faster for me this morning. I guess my machine
varies a lot.  I've now run drefbot and jrefbot back to back and get the
following numbers:
drefbot: 125.192 seconds
jrefbot: 123.916 seconds
That implies the D2 version is 1% slower

boardsize 9
=

genmove b
= E5

ref-nodes
= 111055090

ref-score
= 0.420617

quit
=

user    2m5.328s

--------------

boardsize 9
=

genmove b
= E5

ref-nodes
= 111055090

ref-score
= 0.523835

quit
=

user    2m5.192s


On Fri, 2008-10-24 at 23:09 -0400, Don Dailey wrote:
> Hi Jason,
> 
> That's great!  Thanks.  
> 
> Did you test for basic conformance based on the numbers I reported
> earlier for black score and average nodes per game? 
>  
> 
> I will check it out later if I can get D installed and run the big
> self-test too.   Can you send me a statically compiled linux binary?
> 
> - Don
> 
> 
> On Fri, 2008-10-24 at 23:01 -0400, Jason House wrote:
> > On Sat, 2008-10-18 at 09:26 -0400, Don Dailey wrote:
> > > I have two versions of the reference bot.  A C and a Java version. 
> > 
> > I've ported the Java position.d to D2 and added a basic wrapper around
> > it.  Here are performance numbers on my laptop
> > javabot runs in 132 seconds for 1,000,000 simulations
> > drefbot runs in 146 seconds for 1,000,000 simulations
> > 
> > compilation command with dmd 2.014:
> > dmd -gc -O -release drefbot.d gtp.d position.d phobosrandom.d
> > tangoshuffle.d -ofdrefbot
> > (windows user append .exe to end of line)
> > 
> > Here's a basic description of the files:
> > position.d - A D2 port of position.java
> > gtp.d - GTP implementation using some generic programming
> >         (automatic parameter conversion, uses command registration)
> > drefbot.d - Basic glue of position.d with gtp.d
> > tangoshuffle.d - Used as replacement for Collections.shuffle
> > phobosrandom.d - Patched version of std.random to fix a serious bug
> > 
> > It's possible to eliminate tangoshuffle.d, but I was lazy with my
> > initial port and didn't bother thinking about how to efficiently shuffle
> > (and debug whatever I hacked together)
> > _______________________________________________
> > computer-go mailing list
> > [email protected]
> > http://www.computer-go.org/mailman/listinfo/computer-go/

_______________________________________________
computer-go mailing list
[email protected]
http://www.computer-go.org/mailman/listinfo/computer-go/

Reply via email to