From: Ben Shoemaker <[EMAIL PROTECTED]>

>I 
used 
a 
Palm 
for 
many 
years.  
I 
believe 
Chess 
Genius 
was 
a 
great 
success 
because 
it 
was 
quite 
fast 
(nearly 
instant 
response 
time) 
yet 
it was 
strong 
enough 
to 
give 
the 
average 
user 
a 
good 
game.  
All 
the 
Go 
programs 
I 
played 
on 
the 
Palm 
were 
either 
too 
slow 
and/or 
too weak.  
I 
believe 
AIGO 
for 
Palm 
was 
the 
most 
enjoyable 
overall, 
but 
it 
was 
quite 
weak, 
even 
for 
me 
(I 
am 
only 
18kyu).  
(It 
was 
also 
available in 
a 
Japanese 
language 
version.)  
I 
tried 
OGO, 
but 
it 
was 
incredibly 
slow 
on 
my 
Palm, 
and 
not 
significantly 
stronger 
than 
AIGO.  

>The 
new 
handhelds 
(WindowsMobile/PocketPC, 
Smartphones, 
and 
iPhones) 
can 
all 
run 
versions 
of 
GnuGo 
which 
I 
are 
much 
faster 
and stronger 
than 
anything 
that 
was 
available 
for 
the 
Palm 
platform.  
If 
they 
can't 
already 
support 
the 
latest 
version 
of 
GnuGo, 
they 
will 
soon 

 enough.  
I 
just 
don't 
see 
the 
Palm 
platform 
being 
able 
to 
compete, 
since 
it 
is 
hardly 
used 
anymore.

>Gnugo 
for 
WindowsMobile/PocketPC 
and 
Smartphone: 
http://vieka.com/gnugo/
>Gnugo 
for 
iPhone: 
http://www.robota.nl/products/iPhone%20iGo.html

> I 
say 
keep 
developing 
for 
the 
general 
CPU 
and 
wait 
for 
the 
handheld 
platforms 
to 
catch 
up 
to 
your 
requirements.


That makes sense, considering past programs. But Don Dailey is apparently on 
the trail of a faster and better player for the Palm.
On the one hand, I wonder if the Palm architecture has reached end-of-life. But 
on the other hand, discovering how to improve Go programs for tiny computers 
may lead to design breakthroughs which can also be useful for today's 
multi-gigabyte desktops with dual and quad cores. 

I like where Don is going with the idea of using analysis to create a table of 
position values which could guide playouts and search - it may lead to some 
more widely applicable optimizations, improving the overall quality of play. 

The current study at http://cgos.boardspace.net/study/index.html seems to 
flatten after a certain number of doublings. This may be related to memory 
starvation. Methods which work better on small architectures might also help 
when a gigabyte is not enough. They may suggest ways to optimize the use of 
cache memory on today's CPUs. Algorithms which enable slow processors to play 
adequately may also improve the performance of gigahertz CPUs. 

"Tiny" is relative. Some of my earliest programs were written for a TRS 80 with 
48 K ( yes, kilobytes ) of RAM. I knew a fellow who wrote a Go program back in 
the day for a 64 K Z80-based machine with a 1 MHz ( Megahertz ) clock. It 
wasn't very good, but it was fast enough. 






      
____________________________________________________________________________________
Be a better friend, newshound, and 
know-it-all with Yahoo! Mobile.  Try it now.  
http://mobile.yahoo.com/;_ylt=Ahu06i62sR8HDtDypao8Wcj9tAcJ 

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

Reply via email to