delorie.com/archives/browse.cgi   search  
Mail Archives: djgpp/2002/08/02/08:52:02

From: Hans-Bernhard Broeker <broeker AT physik DOT rwth-aachen DOT de>
Newsgroups: comp.os.msdos.djgpp
Subject: Re: Compiler change
Date: 2 Aug 2002 12:08:56 GMT
Organization: Aachen University of Technology (RWTH)
Lines: 37
Message-ID: <aidsoo$2ke$1@nets3.rz.RWTH-Aachen.DE>
References: <3D4A5576 DOT BFFE13F0 AT earthlink DOT net>
NNTP-Posting-Host: acp3bf.physik.rwth-aachen.de
X-Trace: nets3.rz.RWTH-Aachen.DE 1028290136 2702 137.226.32.75 (2 Aug 2002 12:08:56 GMT)
X-Complaints-To: abuse AT rwth-aachen DOT de
NNTP-Posting-Date: 2 Aug 2002 12:08:56 GMT
Originator: broeker@
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Reply-To: djgpp AT delorie DOT com

Doru Botez <dorub AT earthlink DOT net> wrote:
> We are working on a project in which we use DJGPP for an embedded system
> based on I386Ex and unfortunately the FP emulator EMU387 is very slow,
> so slow that we cannot to process our application in a time frame that
> would allow us to meet our specs. 

Have you tried the other emulator available, WMemu?  If not, I think
you should.  WMemu is essentially the same FPU emulator as used by the
Linux kernel on FPU-less 386 boxes, and it's said to be quite a lot
better than our homegrown emu387.

> without having to change our source code which has to be compatible with
> other platforms. I would like to try to change the compiler ( add
> another switch ) 

Are you sure that switch doesn't already exist?  Have you looked at
'-msoft-float'?

> in order to use the routines provided by QuantASM for
> FP operations. My question is if my idea is feasible in a relatively
> short time ( 1 - 2 weeks ) 

That will depend *very* strongly on the amount of experience you've
had with the GCC and binutils sources.  If you haven't ever looked
into them before, you'll typically need 2 weeks just find your way
through them to the points which you'ld have to change.

Actually, it's quite possible that changing GCC and binutils won't
even be enough --- you'ld have to double-check and optionally rewrite
all libc / libm modules coded in assembly, and all inline assembly in
C modules that use FPU operations to call those other functions.

So, in a nutshell: no, I don't think this is feasible within the given
timeframe.  Not by a wide margin.
-- 
Hans-Bernhard Broeker (broeker AT physik DOT rwth-aachen DOT de)
Even if all the snow were burnt, ashes would remain.

- Raw text -


  webmaster     delorie software   privacy  
  Copyright © 2019   by DJ Delorie     Updated Jul 2019