From: terje.mathisen@nospicedham.tmsw.no   
      
   Rod Pemberton wrote:   
   > On Sat, 10 Nov 2018 12:39:48 +0100   
   > "R.Wieser"  wrote:   
   >   
   >> The problem is that you have degraded the hash to an 8-bit value,   
   >   
   > You really can't "degrade" the original hash much as it's terrible.   
   >   
   > The problem is that ASCII text has insufficient randomness in it's bits   
   > to generate good hash values (i.e., low collisions) from such an   
   > algorithm.  The only solutions are a) increase the randomness of the   
   > input to distribute the hash values more evenly across the hash domain,   
   > or b) increase the size of the hash space, e.g., 16-bit to 32-bit.   
   >   
   > I tested about 50 hash functions with brute-force text input about a   
   > decade ago (comp.lang.misc):   
   >   
   > https://groups.google.com/d/msg/comp.lang.misc/z4dQK8qYhPQ/-5dbeexBcaMJ   
   > (Usenet msg-id hd69d5$umo$1@aioe.org)   
      
   Rod, have you ever looked at the xxHash functions?   
      
   They come in 32 and 64-bit versions, and achieve very high speed due to   
   effectively running 4 hash chains in parallel, with the only real mixing   
   between the chains at the very end. This gets rid of pretty much all the   
   latency problems with classic hash algos.   
      
   Terje   
      
   --   
   -    
   "almost all programming can be viewed as an exercise in caching"   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)