Re: Un-inlining?

pardo@june.cs.washington.edu (David Keppel)
Thu, 9 May 91 00:59:06 GMT

          From comp.compilers

Related articles
Un-inlining? jallen@libserv1.ic.sunysb.edu (Joseph Allen) (1991-05-07)
Re: Un-inlining? pardo@june.cs.washington.edu (1991-05-09)
Re: Un-inlining? preston@titan.rice.edu (1991-05-09)
| List of all articles for this month |

Newsgroups: comp.compilers
From: pardo@june.cs.washington.edu (David Keppel)
Keywords: optimize, design
Organization: Computer Science & Engineering, U. of Washington, Seattle
References: <1991May08.174909.19692@iecc.cambridge.ma.us>
Date: Thu, 9 May 91 00:59:06 GMT

Joseph Allen <jallen@libserv1.ic.sunysb.edu> writes:
>Do any optimizers un-inline (outline)?


Yes, it's called `proceduralization'.


The idea is to take common idioms and stuff them in to a single
function.


See the Irvine Program Transformation Catalogue and


%A Christopher W. Fraser
%A Eugene W. Myers
%A Alan L. Wendt
%T Analyzing and Compressing Assembly Code
%B Proceedings of the ACM SIGPLAN '84 Symposium on Compiler
Construction
%J SIGPLAN Notices
%V 19
%N 6
%D June 1984
%P 117-121




A related operation is to optimize the common path through a function.
Possible optimizations include delaying register spill/restore until the
values are actually used (rather than ``well they're used *somewhere* in
the function; consider the SPARC `save' and `restore' instructions) and
stack adjusts for storage that is declared at the outermost level but is
used only at some inner level(s).


I know of no compilers that do the latter.


;-D on ( In Unlining? ) Pardo
--


Post a followup to this message

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