[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [computer-go] Pattern Matcher



Don,

I don't think what you wrote can be processed by Vincent's mind.  He is now stuck between the
proverbial rock and a hard place - the rock (his taking back his over-assertions regarding Java)
and a hard place (proving he can really optimize C in comparison to optimizing Java to support his
over-assertions).

Honestly, I think we can all pretty clearly see the gauntlet was thrown (by John, I guess).  And
we can also see, it isn't being picked up by Vincent (at least in a non BS way).  We'll see if
Vincent will eventually really pick it up and demonstrate his "knowledge".  Until then, cat's out
of the bag, so to speak.  ;^)

BTW, even this test is not optimal use of Java.  It is just how fast can Java code, crafted to
look identical to C code, can execute.  I understand that was the point of this apples-to-apples
test.

However, the Java JVMs (both Sun's and IBMs) use optimizations that assume good OO designed and
implemented code.  The execution prediction, out of order instruction execution. CPU register and
cache management (including per processor type) and memory management strategies in these JVMs
work even more effectively with proper OO use (contrary to Vincent's less-informed and clearly
strongly biased opinions).


Jim


--- Don Dailey <drd@xxxxxxxxxxxxxxxxx> wrote:

> 
> Vincent, your  speedup has  taken the solver  over 10 times  longer to
> compute!  How can you call that a speedup?  If you turn off alpha/beta
> pruning it will go even faster!
> 
> You are embarassing yourself and the  rest of us chess guys.  Make the
> program SOLVE faster and check your work before reporting progress.
> 
> - Don
> _______________________________________________
> computer-go mailing list
> computer-go@xxxxxxxxxxxxxxxxx
> http://www.computer-go.org/mailman/listinfo/computer-go/
> 


_______________________________________________
computer-go mailing list
computer-go@xxxxxxxxxxxxxxxxx
http://www.computer-go.org/mailman/listinfo/computer-go/