#4096: New primops for indexing: index*OffAddrUsing# etc
---------------------------------+------------------------------------------
    Reporter:  simonpj           |        Owner:              
        Type:  feature request   |       Status:  new         
    Priority:  normal            |    Milestone:  6.16.1      
   Component:  Compiler          |      Version:  6.12.2      
    Keywords:                    |   Difficulty:              
          Os:  Unknown/Multiple  |     Testcase:              
Architecture:  Unknown/Multiple  |      Failure:  None/Unknown
---------------------------------+------------------------------------------

Comment(by rl):

 I implemented this and it gives very nice speedups (up to 1.5) for the
 vector benchmarks. The patch is attached. It has one big problem, though.
 The new primops must be ok-for-speculation regardless of their first
 argument (the boxed term that is being touched). I hacked this by
 explicitly testing for these primops (all 16 of them) in
 `exprOkForSpeculation` but that's really just a hack. I'm not sure how to
 handle this properly. The situation is somewhat similar to division
 primops which are also hardwired in `exprOkForSpeculation`.

-- 
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/4096#comment:2>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler
_______________________________________________
Glasgow-haskell-bugs mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs

Reply via email to