home bbs files messages ]

Forums before death by AOL, social media and spammers... "We can't have nice things"

   comp.compilers      Compiler construction, theory, etc. (Mod      2,753 messages   

[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]

   Message 883 of 2,753   
   Karsten Nyblad to All   
   Re: Jump size optimization info...   
   28 Jan 07 01:41:31   
   
   From: 148f3wg02@sneakemail.com   
      
   > [To get the genreral NP complete subtraction problem I'd think you   
   > could add branch chaining, e.g., if you have A->C and B->C, you can   
   > change that to A->B and B->C.  -John]   
      
   A way of making the problem NP-complete is to allow reordering of code.   
      
     In some cases you can save branches or save long branches by   
   reordering the code, e.g., by swapping the order of two basic blocks.   
   Unfortunately the CACM paper John referenced to proves that that is   
   NP-complete.  The paper goes on to prove that the algorithm described   
   by Steven Nichols is optimal as long as you are forbidden to reorder   
   the code.   
      
   Karsten Nyblad   
   148f3wg02 at sneakemail dot com   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)   

[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]


(c) 1994,  bbs@darkrealms.ca