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

Re: computer-go: Computer Go hardware



Yes,

My intuition is that intelligence, or the  simulation of it requires a
lot more that  incredibly sophisticated programming, although  that is
absolutely necessary too.  Like the  human brain, it requires enormous
processing power  (the human brain may be  the  most sophisticated and
powerful computing device we know of currently.)  Currently, the human
brain still plays the best  Go and probably will  for a very long time
to come.

We have seen things happen in the simulation of intelligence that were
completely unthinkable a few years ago, due to the fact that computers
have  finally become powerful  enough to think  about the unthinkable.
Speech recognition is  one thing that comes to  mind.  The human brain
handles this task with ease, and still very much better.

In the chess world, there have been many discussion (arguments really)
over 2 styles of thinking about computer  chess.  There was the "brute
force" camp VS the "intelligent"  approach.  The brute force advocates
viewed themselves as pragmatic engineers and the "slow but smart" camp
viewed themselves as creative intellectual geniuses.  To me, they both
seemed  to be  snobs  in their own  way  but meanwhile   the very best
programmers   were busy writing  successful  programs that were ALWAYS
both very  smart and very fast.  Yes,  some had more  of  one than the
other,  but  no "stupid" program did   that well and  neither  did any
"slow" programs.

I can't believe that  we are doing things  correctly if we  don't know
how  to  exploit  extra  hardware  resources.   I  like your   idea of
exploring how parallelism might help.   

Don



   From: Bill Rankin <billran@xxxxxxxxxxxxxxxxx>
   Date: Sun, 1 Apr 2001 11:07:56 -0400 
   MIME-Version: 1.0
   X-Mailer: Internet Mail Service (5.5.2650.21)
   Sender: owner-computer-go@xxxxxxxxxxxxxxxxx
   Precedence: bulk
   Reply-To: computer-go@xxxxxxxxxxxxxxxxx
   Content-Type: multipart/alternative;
	   boundary="----_=_NextPart_001_01C0BABD.885A2BD0"
   Content-Length: 4569

   This message is in MIME format. Since your mail reader does not understand
   this format, some or all of this message may not be legible.

   ------_=_NextPart_001_01C0BABD.885A2BD0
   Content-Type: text/plain

   Interesting rant.  Was there any purpose behind this posting other than
   to generate random discussions?

   Another approach worth pursuing is 
   to connect many cpus together. Not for parallel processing in the 
   conventional sense. Instead connect them in a certain topological way. 


   Firstoff, what you describe *is* conventional parallel processing.  Grids,
   hyper-cubes, fat-trees, et. al., has all been been extensively researched.
   If you can define a regular topology the is useful and isn't a combination
   of the 
   more established sets, then please let me know and we can apply for some
   research funding :-)

   Other than that, I have been looking with some interest at the discussion of
   the performance bottlenecks in the GnuGo code and was wondering if a 
   beowulf/cluster type of approach could be used to improve things a bit?

   Anyone have any thoughts on this matter?

   -bill rankin

   ------_=_NextPart_001_01C0BABD.885A2BD0
   Content-Type: text/html

   <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
   <HTML><HEAD>
   <META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=us-ascii">


   <META content="MSHTML 5.00.3019.2500" name=GENERATOR></HEAD>
   <BODY>
   <DIV><FONT color=#0000ff face=Arial size=2><SPAN 
   class=972455914-01042001>Interesting rant.&nbsp; Was there any purpose behind 
   this posting other than</SPAN></FONT></DIV>
   <DIV><FONT color=#0000ff face=Arial size=2><SPAN class=972455914-01042001>to 
   generate random discussions?</SPAN></FONT></DIV>
   <BLOCKQUOTE 
   style="BORDER-LEFT: #0000ff 2px solid; MARGIN-LEFT: 5px; PADDING-LEFT: 5px">
     <DIV><FONT size=2><FONT face=Arial>Another approach worth pursuing is <BR>to 
     connect many cpus together. Not for parallel processing in the 
     <BR>conventional sense. Instead connect them in a certain topological 
     way.&nbsp;<BR><FONT color=#0000ff><SPAN 
     class=972455914-01042001>&nbsp;</SPAN></FONT></FONT></FONT><FONT size=2><FONT 
     face=Arial><FONT color=#0000ff><SPAN 
     class=972455914-01042001>&nbsp;</SPAN></FONT></FONT></FONT></DIV></BLOCKQUOTE>
   <DIV style="MARGIN-RIGHT: 0px"><FONT size=2><FONT face=Arial><FONT 
   color=#0000ff><SPAN class=972455914-01042001>Firstoff, what you describe *is* 
   conventional parallel processing.&nbsp; Grids,</SPAN></FONT></FONT></FONT></DIV>
   <DIV style="MARGIN-RIGHT: 0px"><FONT size=2><FONT face=Arial><FONT 
   color=#0000ff><SPAN class=972455914-01042001>hyper-cubes, fat-trees, et. al., 
   has all been been extensively researched.</SPAN></FONT></FONT></FONT></DIV>
   <DIV style="MARGIN-RIGHT: 0px"><FONT size=2><FONT face=Arial><FONT 
   color=#0000ff><SPAN class=972455914-01042001>If you can define a regular 
   topology the is useful and isn't a combination of the 
   </SPAN></FONT></FONT></FONT></DIV>
   <DIV style="MARGIN-RIGHT: 0px"><FONT size=2><FONT face=Arial><FONT 
   color=#0000ff><SPAN class=972455914-01042001>more established sets, then please 
   let me know and we can apply for some</SPAN></FONT></FONT></FONT></DIV>
   <DIV style="MARGIN-RIGHT: 0px"><FONT size=2><FONT face=Arial><FONT 
   color=#0000ff><SPAN class=972455914-01042001>research funding 
   :-)</SPAN></FONT></FONT></FONT></DIV>
   <DIV style="MARGIN-RIGHT: 0px"><FONT size=2><FONT face=Arial><FONT 
   color=#0000ff><SPAN 
   class=972455914-01042001></SPAN></FONT></FONT></FONT>&nbsp;</DIV>
   <DIV style="MARGIN-RIGHT: 0px"><FONT color=#0000ff face=Arial size=2><SPAN 
   class=972455914-01042001>Other than that, I have been looking with some interest 
   at the discussion of</SPAN></FONT></DIV>
   <DIV style="MARGIN-RIGHT: 0px"><FONT color=#0000ff face=Arial size=2><SPAN 
   class=972455914-01042001>the performance bottlenecks in the GnuGo code and was 
   wondering if a </SPAN></FONT></DIV>
   <DIV style="MARGIN-RIGHT: 0px"><FONT color=#0000ff face=Arial size=2><SPAN 
   class=972455914-01042001>beowulf/cluster type of approach could be used to 
   improve things a bit?</SPAN></FONT></DIV>
   <DIV style="MARGIN-RIGHT: 0px">&nbsp;</DIV>
   <DIV style="MARGIN-RIGHT: 0px"><FONT color=#0000ff face=Arial size=2><SPAN 
   class=972455914-01042001>Anyone have any thoughts on this 
   matter?</SPAN></FONT></DIV>
   <DIV style="MARGIN-RIGHT: 0px"><FONT color=#0000ff face=Arial size=2><SPAN 
   class=972455914-01042001></SPAN></FONT>&nbsp;</DIV>
   <DIV style="MARGIN-RIGHT: 0px"><FONT color=#0000ff face=Arial size=2><SPAN 
   class=972455914-01042001>-bill rankin</SPAN></FONT></DIV></BODY></HTML>

   ------_=_NextPart_001_01C0BABD.885A2BD0--