[11217] in Athena Bugs
NetProb ticket #639, decmips 7.6G: cygnus gcc (/mit/cygnus/decmipsbin/gcc)
daemon@ATHENA.MIT.EDU (probe@MIT.EDU)
Fri Oct 15 15:14:53 1993
From: probe@MIT.EDU
Date: Fri, 15 Oct 1993 15:14:46 -0400
To: reidmp@MIT.EDU
Cc: bugs@MIT.EDU
I have you listed as the person who deals with most of the Cygnus software
questions. Offhand, I would suspect that this will need to be forwarded to
Cygnus once a code example has been obtained.
-Richard
>From tlc@MIT.EDU Sat Oct 9 18:17:24 1993
>Received: by gregor.MIT.EDU (5.57/4.7) id AA29703; Sat, 9 Oct 93 18:17:23 -040>0
>Received: from W20-575-79.MIT.EDU by MIT.EDU with SMTP
> id AA19315; Sat, 9 Oct 93 18:17:07 EDT
>From: tlc@MIT.EDU
>Received: by w20-575-79.MIT.EDU (5.57/4.7) id AA06157; Sat, 9 Oct 93 18:17:06 >-0400
>Message-Id: <9310092217.AA06157@w20-575-79.MIT.EDU>
>To: bugs@MIT.EDU
>Subject: decmips 7.6G: cygnus gcc (/mit/cygnus/decmipsbin/gcc)
>Date: Sat, 09 Oct 93 18:17:05 EDT
>
>System name: w20-575-79
>Type and version: KN02ba 7.6G
>Display type: PMAGB-BA
>
>What were you trying to do?
>Writing a fast counting sort procedure.
>
>What's wrong:
>Used gcc with -O2, and the sort occured correctly. Used
>gcc with -O2 -funroll-loops, and the numbers were no
>longer sorted despite no changes to the code.
>
>What should have happened:
>The list should have been sorted correctly -- I believe my code is
>correct, and the fact that it functioned with -O2 leads me to
>believe that there is something wrong with -funroll-loops.
>
>Please describe any relevant documentation references:
>The file which I tried to compile is available as
>/mit/tlc/misc/courses/6.046/one/one.c -- the relevant function is
>sort_words.
>
>
>I hope this helps.
>
>max
>
