From: tkoenig@netcologne.de   
      
   Stefan Monnier  schrieb:   
   > BTW, when discussing ISA compactness, I usually see it measured by   
   > comparing the size of the code segment in typical executables.   
   >   
   > I understand that it's as good a measure as any and it's one that's   
   > fairly easily available, but at the same time it's not necessarily one   
   > that actually matters since I expect that it affects a usually fairly   
   > small proportion of the total ROM/Flash/disk space.   
   >   
   > I wonder if there have been other studies to explore other impacts such   
   > as run time, or cache miss rate.   
      
   Optimizing compilers also play a large role.  Very many   
   optimizations increase code size.  A few examples include loop   
   unrolling, aligning, inlining and function cloning.  (For those   
   not familiar with the term: Assume you have a function that is   
   called with constant arguments; the compiler might want to make a   
   copy of that function with a certain constant parameters, if this   
   can lead to simplification and thus further optimization of code   
   on a hot path).   
      
   The resulting code size increases might then cause problems with   
   icache pressure, which is why some people report faster execution   
   with lower optimization levels with some code.   
      
   I'm not aware of any compiler which has an icache model, but this   
   would also be very hard (presumably).   
      
   The only fair comparision, I think, is with -Os.   
      
   --   
   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)