From: khan AT xraylith DOT wisc DOT edu (Mumit Khan) Subject: Re: Maths ruined when optimizing in gcc 25 Nov 1996 20:58:46 -0800 Sender: daemon AT cygnus DOT com Approved: cygnus DOT gnu-win32 AT cygnus DOT com Distribution: cygnus Message-ID: <9611260416.AA08499.cygnus.gnu-win32@modi.xraylith.wisc.edu> Original-To: setzer AT rsdws3 DOT herl DOT epa DOT gov Original-Cc: "Pedro A. Aranda =?iso-8859-1?Q?Guti=E9rrez?=" , gnu-win32 AT cygnus DOT com In-Reply-To: Your message of "Mon, 18 Nov 1996 10:09:27 EST." Original-Sender: owner-gnu-win32 AT cygnus DOT com Woodrow Setzer writes: > On Fri, 15 Nov 1996, Pedro A. Aranda [iso-8859-1] Guti=E9rrez wrote: > > > ... > > When I compile WITHOUT optimization, floating point > > ops are ok. If I dare do any optimization, I always > > get all sorts of errors. > >=20 > > Has anyone experieced a similar thing? > > This may be related. I am working with an f2c-converted version of > dmngb.f, one of the routines for optimization from the ATT PORT library.=20 > It compiles and works OK using gcc 2.7.2.1 under Solaris 2.3 on a Sun > Sparc 20 using the option -O. Under cyg-win32 and the same option, the > code compiles and runs without compile-time or run-time errors, but I get > the wrong answer. If I omit the -O flag on the compiler invocation under > cyg-win32, I get essentially the same answer as I do on the Sun.=20 > > -- Another data point here. I'm in the process of testing g77 port to cygwin32 (it was quite simple once I figured out the right flags to libfI77) and the big gotcha is the every time I use -O, some of the programs produce NaNs all over the place! Haven't been able to pin point where it happens yet, but hopefully soon. Anybody have a short program that screws up the numerics with -O? If so, please email it me (if fortran, then either original or f2c'd will do fine). regards, Mumit -- khan AT xraylith DOT wisc DOT edu http://www.xraylith.wisc.edu/~khan/ - For help on using this list, send a message to "gnu-win32-request AT cygnus DOT com" with one line of text: "help".