Date: Fri, 2 Feb 1996 17:52:05 -0800 (GMT)
From: Orlando Andico <oandico AT eee DOT upd DOT edu DOT ph>
To: Terence Abbott <t DOT s DOT abbott AT larc DOT nasa DOT gov>
cc: djgpp AT delorie DOT com
Subject: Re: Speed,V1&V2,Pentium&486
Message-ID: <Pine.SOL.3.91.960202174500.3863A-100000@gollum>
MIME-Version: 1.0

On 1 Feb 1996, Terence Abbott wrote:

> Here's another interesting note.  I've got a real-time
> graphics program (~65k lines of source code) that I've
> compiled and executed with the following results:
>   DJGPP 1.12m4 with GRX 1.03 :  18 iterations/sec.
>   DJGPP 2.0beta4 with GRX 2.0:  8 iterations/sec.
> All on a 486/66.

That's odd. Isn't GCC 2.7.2 supposed to have more optimizations builtin? 
I haven't got 2.0beta4; beta3 only has GCC 2.6.3. Does beta4 use the same 
compiler? I also read DJ's post about beta5... *is* this the final beta? 
does it have GCC 2.7.2? I'm working on a very computational-hungry image 
processing application; it does very little I/O so switching to RM 
shouldn't be too common. But I haven't compared any speed changes from 
1.12m4 and 2.0beta3. Will they be _that_ drastic?

Also, in the GCC 2.7.2 tardist, I see an "ms-dos" directory in there, 
plus some instructions for cross-compilation. Will this let me build GCC 
2.7.2 on a Linux box with DJGPP output? of course, the libs which come 
with DJGPP have to be used... I know there's something in the Massive FAQ 
about this, but that was for an older version of GCC. It seems that in 
the 2.7.2 distribution the FSF people already put in the hooks for people 
using DJGPP. Is this right?

/----------------------------------------------------------------------------\
|  Orlando A. Andico                 "I have no concept of time, other than  |
|  oandico AT eee DOT upd DOT edu DOT ph               it is flying." -- Alanis Morissette  |
\----------------------------------------------------------------------------/