Re: [cc65] Optimization for size

From: Andrew J. Kroll <forge1dr.ea.ms>
Date: 2006-04-05 12:11:31
> 
> 
> On Wed, Apr 05, 2006 at 05:57:56AM +0200, Groepaz wrote:
> > that sounds pretty extreme... are you sure you had the compiler optimization
> > enabled? (also, did you check ftp://ftp.musoftware.de/pub/groepaz/opt65.c ?)
> 
> I could not test the thing because of flex constraints ("input rules are too
> complicated (>= 32000 NFA states)"), but I had a look at it. What Andrew does
> is to replace common blocks by subroutines. That works well with larger
> programs. The flex rules look like they are targeted to a specific program, so
> other programs won't benefit as much. Which is not a problem if you generate
> new rules specific for a program.
> 

Wich version of flex? I use 2.5.33 on debian testing, with the -Ca option...

As far as rule generation, the program to create the rules is actualy a 
complex set of custom programs that recognize patterns in text, while
rejecting things that are potentially "bad".

JSR/JMP optimizations to a library is nothing new... infact...
the libc is exactly that... :-)

Yours,
-- 
Andrew J. Kroll
Software and Hardware Engineer and Consultant
Grafixoft http://dr.ea.ms/HW/
----------------------------------------------------------------------
To unsubscribe from the list send mail to majordomo@musoftware.de with
the string "unsubscribe cc65" in the body(!) of the mail.
Received on Wed Apr 5 12:11:41 2006

This archive was generated by hypermail 2.1.8 : 2006-04-05 12:11:44 CEST