From: ademakov@nospicedham.gmail.com   
      
   Thank you!   
      
   On Sunday, May 12, 2019 at 8:05:03 PM UTC+3, Anton Ertl wrote:   
   > What I notice is that the benchmark does all iterations with the same   
   > alignment (which may be optimal) and the same size (so branch   
   > prediction will be optimal); these settings may not be representative.   
   >   
      
   I agree with you. Perhaps I need to add test modes with different sizes and   
   alignment values.   
      
   > I also notice that the Haswell seems to have two different kinds of   
   > performance results:   
      
   Hmm, the results might depend on the background activity and on cpu energy   
   saving mode.   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)