RE: [computer-go] Java speed vs.C speed

["Mark Boon" <tesuji@xxxxxxxxx>]

> -----Original Message-----
> From: computer-go-bounces@xxxxxxxxxxxxxxxxx
> [mailto:computer-go-bounces@xxxxxxxxxxxxxxxxx]On Behalf Of Frank de Groot
> Sent: Tuesday, November 09, 2004 14:01
> To: computer-go
> Subject: Re: [computer-go] Pattern Matcher
>
>
> From: "Marco Scheurer" <marco@xxxxxxxxxxxxxxxxx>
> To: "computer-go" <computer-go@xxxxxxxxxxxxxxxxx>
> Subject: Re: [computer-go] Pattern Matcher
>
>
>
> > Yeah... I found this thread interesting for a while and appreciated to
> > see the input from chess programmers. However, a discussion about the
> > relative merits of C and Java is definitely off-topic and computer
> > languages have already been discussed at length on this list.
>
> Then again, it would be a disaster to shut up the world's most successful
> (and willing to share!) comp. chess experts here.
>
> Those few people that are not interested enough in all aspects of comp. Go
> to endure some message floods occasionally, should adjust the list
> subscription to "daily digest" or "weekly digest" if that exists.
>
> I am *extremely* interested in Java vs. C speed topics because I thought,
> like Vincent, that is was very slow, and that was the only reason I am not
> using it now, which is a very sad thing, as I am not not cross-OS
> compatible.
>
> I have never, ever seen a refutation of the "Java is much slower than
> C/Delphi" so if Vincent & Mark can shed light on this and solve the issue
> once and for all, I might even start porting to Java.
>

I agree that the subject is of interest, but I think the discussion has
taken a stupid turn and people rightly complain about it.

I'm saying I see only a minor (I consider 30-50% minor) performance hit for
Goliath in Java. What more light do you want me to shed? You don't really
think I'm going to publish the source-code for Goliath to make this point,
do you?

John is showing an example that shows the same thing. Instead of saying
"interesting, maybe there's a point there" you guys complain about the
algorithm. But the algorithm is not the point, it's whether Java runs it
almost as fast as C does or not, and whether the algorithm uses programming
constructs that are used in Go. If you don't agree with the algorithm, come
up with your own example, and show us the code, the numbers, whatever. But
you'll have to stay relevant for computer-Go, because anything else is not
of interest to us.

We're showing programming constructs that are relevant to Go and show that
Java runs these almost as fast as C does. To me that *proves*, as best as
you can prove in a case like this. Now, if you guys are saying: "but these
constructs can be greatly optimized!", then we should listen because maybe
we can use your advice. But until these 'optimized' constructs have proven
to be beneficial to Go too (and don't cause development to slow down to a
crawl) then it is beside the point.

When your constructs prove to be indeed beneficial to computer-Go then we
can implement them as well, and redo the comparison. Until that happens I
think you'll have to accept what the numbers are showing now.


> Saying: "ploink" and "shut up" is a bad idea.
> Those guys are computer-game guru's and everything they say might be worth
> gold or more.
>
> They are arrogant bastards but *highly* successful arrogant bastards :)
>
> _______________________________________________
> 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/