Re: Data Structure Reorganizing Optimizations

glew@ichips.intel.com (Andy Glew) writes:


>I've been meaning to post a summary of the discussion to this group.
>Will still go into more detail. But the short of it is:


>Problems with reorganization mainly lie in people making assumptions
>about the *consistency* of data structure layout.


[Remainder deleted]




Rather than listing a bunch of exceptions where the optimizations won't
take place, with the inevitable cases where the compiler optimized something
it shouldn't have, why not use a C keyword to indicate when the optimizations
*can* be done? This allows the optimizations to be done, without causing
any problems for older code. Unfortunately, a good keyword doesn't exist,
but we can pick one from the list of reserved words. Maybe a 'void struct'?
Or how about that old 'entry' keyword that's never been used? :-) It would
not make much sense to see an 'entry struct' and know that optimization can
be applied, but on the whole I'd rather see something like that done rather
than have a compiler attempt to deduce when it can reorder my structures
and when it can't on its own!


--
Doug Siebert
dsiebert@isca.uiowa.edu
--

Return to the comp.compilers page.
Search the comp.compilers archives again.