Re: Loop clean up fixes

[Michael Hayes <m dot hayes at elec dot canterbury dot ac dot nz>]

Alexandre Oliva writes:
 > On Jan 19, 2000, Michael Hayes <m.hayes@elec.canterbury.ac.nz> wrote:
 > 
 > > Anyway, it's only an interim measure since my low-overhead looping
 > > patches does away with the used_count_register field altogether.
 > 
 > In this case, wouldn't it be better to use my patch instead?

In retrospect, yes it would have been a more pragmatic short-term
solution.  Late last night I convinced myself that it would be a good
idea to retain the loop iteration info for all the loops of a function
so that we could perform better loop unrolling and loop fusion; rather
than just keeping a single copy for the current loop being processed.
Now I'm not sure.  What do others think?  Would this be worth the
additional few dozen bytes/loop?

Michael.