gnupic: Re: [gnupic] Linker Optimizations


Previous by date: 2 Mar 2007 21:51:05 +0000 Re: [gnupic] Linker Optimizations, Scott Dattalo
Next by date: 2 Mar 2007 21:51:05 +0000 [PIC] Piklab 0.14.0, Nicolas Hadacek
Previous in thread: 2 Mar 2007 21:51:05 +0000 Re: [gnupic] Linker Optimizations, Scott Dattalo
Next in thread: 2 Mar 2007 21:51:05 +0000 Re: [gnupic] Linker Optimizations, Scott Dattalo

Subject: Re: [gnupic] Linker Optimizations
From: "David Barnett" ####@####.####
Date: 2 Mar 2007 21:51:05 +0000
Message-Id: <0b0001c75d14$8667ff00$2001a8c0@barnett2>

----- Original Message ----- 
From: "Scott Dattalo" ####@####.####
To: ####@####.####
Sent: Friday, March 02, 2007 3:14 PM
Subject: Re: [gnupic] Linker Optimizations


> On Fri, 2007-03-02 at 12:47 -0600, David Barnett wrote:
>>  I've always been frustrated about some big optimizations that can only
>>  be made manually; the linker doesn't seem to have enough information
>>  to optimize them.  HI-TECH will avoid these issues by using
>>  "omnicode", meaning all source files are compiled at the same time,
>>  and SourceBoost seems to skirt the problem with a custom object file
>>  format.  The two that stand out to me are:
>>  a) code-flow-related optimizations on page and bank selections
>>  b) removal of unused functions and global variables
>>
>
> <snip>
>
> David,
>
> I agree with you fully (for whatever that's worth). In my opinion, the
> linker should be capable of Bank selection, Simulator assertions for
> banks, tail optimization, dead code elimination, peephole optimization,
> etc. and much more.
Well, I'm glad to hear you agree.  I figured this wouldn't be a priority for 
anyone else.

> I suggested at one point that the pCode optimizer I started in SDCC should
> be ported into gputils. In it's first implementation, it did much of the
> stuff we'd want. The pCode optimizer was written before the linker was
> used too much. So consequently, it dealt with single C-files at a time.
> When linker support became more prevalent, the pCode optimizer had to be
> changed quite a bit.
This sounds very interesting and ambitious.  I'm not quite sure what 
constitutes pCode, though.  For SDCC, that's what gets passed from the front 
end to the back end, right?  I ask because I wasn't aware you could extract 
pCode from pure assembly code.  As I mentioned in the last email, there's 
some flow information that you could extract from C, but I don't think from 
asm, that would be necessary for a lot of the optimizations.

I'd love to see any information you have about that suggestion to port the 
pCode optimizer.  It looks like gputils may be the best place for me to get 
started helping out.  My biggest ambition is to make some radical 
improvements to SDCC, eventually.

> The only issue I have is time...
And nobody blames you there.  You've put a ton of time into gnupic projects 
(Thanks!).  What you need is more of other people's time =).

David 


Previous by date: 2 Mar 2007 21:51:05 +0000 Re: [gnupic] Linker Optimizations, Scott Dattalo
Next by date: 2 Mar 2007 21:51:05 +0000 [PIC] Piklab 0.14.0, Nicolas Hadacek
Previous in thread: 2 Mar 2007 21:51:05 +0000 Re: [gnupic] Linker Optimizations, Scott Dattalo
Next in thread: 2 Mar 2007 21:51:05 +0000 Re: [gnupic] Linker Optimizations, Scott Dattalo


Powered by ezmlm-browse 0.20.