From: oml1 AT Ra DOT MsState DOT Edu (Owen LaGarde)
Newsgroups: comp.os.msdos.djgpp
Subject: Re: DJGPP 2.01 + OpenDOS = total ... hello ...
Date: 8 May 1997 15:29:16 GMT
Organization: Mississippi State University
Lines: 21
Message-ID: <5ksrgc$ote$1@NNTP.MsState.Edu>
References: <Pine DOT SUN DOT 3 DOT 91 DOT 970504140635 DOT 9215Y-100000 AT is> <5kkpmn$agg$1 AT NNTP DOT MsState DOT Edu> <5kqq5c$tun$1 AT news DOT inconnect DOT com>
NNTP-Posting-Host: ra.msstate.edu
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Precedence: bulk

    Here's an interesting update ... I pulled down a new copy, but as the
install-from-dir instead of install-from-disk.  THAT worked fine with
DPMS loaded, DPMI off, no framing, no exclusions, 6M norton cache and 6M
temp dir as the root of a vdisk.  Libgrx20.a spun up fine and the test
series ran like a charm, though a few other commercial dos4gw apps didn't
like the lack of extended or the dpms at all.  A boot menu solves that for
now.  The disk install set still faults on the very same config ... hmmm.
Me get suspicious, disks bad! 
    On a side note ... having never used the djgpp gcc, how much of a
difference is there in execution speed between -g, none, and -O parms?
I took the "life" grx20 test prog and pulled in the stub code that was 
stuck off in includes for use with compiling the entire test set, and
it compiled (-O) to exactly the same size (I don't have a crc at the moment)
but runs easily 1/3 as fast as the original.  I've reproduced the gcc calls
(just trimmed the makefile) so I don't THINK I've missed anything... what
else could account for this?
-- 
 Owen LaGarde                           |     Performance prediction through
 Forest Products Research Laboratory    |       AI-driven process simulation
 Mississippi State University           | 
 oml1 AT ra DOT msstate DOT edu (130.18.80.10)     |   "... Oh, what a Tangled Web ..."