From: erayo@bilkent.edu.tr   
      
   Robert Bruce Carleton  wrote in message news:<   
   10853d6$1@news.unimelb.edu.au>...   
   > I've been reading the "The Connection Machine" by Danny Hillis and as   
   > a result found the Teranex GAPP while researching SIMD programming.   
   > The GAPP is a geometric-arithmetic parallel processor that was   
   > patented by Lockheed Martin.  It's now being used in video processing.   
   >   
   > I'm not in the AI field so my question might be kind of naive.  Has   
   > any one looked at using the GAPP processor for AI work?   
   >   
   > I don't work for Teranex or one of their competitors.  I'm a curious   
   > amateur that wonders if SIMD programming techniques are still being   
   > used in AI, or if it's been surpassed by other techniques or   
   > technologies.   
      
   I think it is much more cost effective to use a Beowulf cluster for   
   parallel processing in general. However, as for SIMD, most   
   computationally heavy AI systems like evolutionary programming would   
   be better off using an MIMD architecture. The SIMD architecture is   
   excellent for vector operations and the like (and fitting for DSP work   
   perhaps), but I doubt it is useful for data mining, machine learning,   
   evolutionary programming, or any other subfield of AI that might   
   require huge processing power. In particular, data mining can greatly   
   benefit from parallel processing due to massive datasets, but the   
   task/data dependency is not adequately captured by SIMD model, even   
   for the most "straightforward" mining tasks (such as Association Rule   
   Mining, but mind you ARM is not straightforward at all!).   
      
   Regards,   
      
   --   
   Eray Ozkural   
      
   [ comp.ai is moderated.  To submit, just post and be patient, or if ]   
   [ that fails mail your article to , and ]   
   [ ask your news administrator to fix the problems with your system. ]   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)