From: tkoenig@netcologne.de   
      
   (This might be blindingly obvious to most regulars, but I thought   
   I'd post this, just in case for some discussion)   
      
   SIMD is not always about vectorizing loops, they can also be used   
   for tree-shaped reductions (not sure what the canonical name is).   
      
   Consider the following problem:  You have 128 consecutive bytes and   
   want to find the minimum value, and you have 512-bit SIMD registers.   
      
   You load the values into two 512-bit = 64 byte registers A and B, into   
   positions A<0:7>, A<8:15>, ... A<504:511>, and B correspondingyl.   
      
   You do a SIMD 8-bit "min" operation.  The target c register contains   
   the minimum values: min(A<0:7>,B<0:7>), min(A<8:15>,B<8:15>, ...   
   min(A<504:511>,b<504:511>).  C need not be distinct from A or B.   
   You then have 64 values.   
      
   You then move the upper values of C into register D (which need not   
   be distinct from A or B), giving you D<0:7=min(A<256:263>,B<256:263)   
   etc.   
      
   You then do the min operation with half the length of the registers,   
   giving you 32 values.   
      
   And so on, until you have the single value, which is reached in   
   seven steps.   
      
   This kind of thing is, AFAIK, used in high-performance code such   
   as JSON parsers or regexp matchers.  An example (in principle)   
   can be found at https://gitlab.ethz.ch/extra_projects/fastjson/ .   
      
   Just wondering... is this somethig that VVM or similar could   
   also do? Or does this actually require SIMD and the necessary   
   shift, rotate or permute intructions?   
   --   
   This USENET posting was made without artificial intelligence,   
   artificial impertinence, artificial arrogance, artificial stupidity,   
   artificial flavorings or artificial colorants.   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)