|Optimizing structure layout email@example.com (1997-03-21)|
|Re: Optimizing structure layout firstname.lastname@example.org (Julian Dolby) (1997-03-22)|
|Re: Optimizing structure layout email@example.com.OZ.AU (1997-03-22)|
|Re: Optimizing structure layout firstname.lastname@example.org (Christopher Glaeser) (1997-03-22)|
|Re: Optimizing structure layout email@example.com (Michael Meissner) (1997-03-27)|
|Re: Optimizing structure layout firstname.lastname@example.org (David L Moore) (1997-03-27)|
|Re: Optimizing structure layout email@example.com (1997-03-31)|
|[2 later articles]|
|Date:||21 Mar 1997 10:24:58 -0500|
|Keywords:||optimize, storage, question, comment|
While it is quite common that compilers optimize the code they
produce, I haven't heard of a commonly used system that really
optimizes the layout of the data structures that are generated. Are
there such systems?
What I have heard of is reordering the fields and/or inserting some
padding in order to comply to alignment constraints. But what about
omitting fields that are never used; what about omitting fields that
can be trivially computed from the other fields; what about inserting
"caching" fields that hold values that are expensive to compute?
Such a system necessarily needs some sort of access control to the
data structure, i. e. "privacy" or "information hiding". It probably
also needs profiling information to detect cases where unneeded
information is calculated or where information is calculated too
Does lazy evaluation go into this direction?
Erik Schnetter, firstname.lastname@example.org
[I believe that you'll find the data base crowd think about optimizations
like this, since they tend to have higher level data descriptions to work
from than most programming languages do. -John]
Return to the
Search the comp.compilers archives again.