On Thu, Apr 19, 2001 at 10:50:34AM +0100, Nick Bailey wrote: > Ionut Georgescu wrote: > > > PS command line: g77 -g -pg -o programm *.f > > PPS gcc -v: > > Reading specs from /usr/lib/gcc-lib/alpha-linux/2.95.3/specs > > gcc version 2.95.3 20010315 (Debian release) > > I'm interested... I don't know enough FORTRAN to test this out, but I have > been > seeing Scilab (which has a lot of FORTRAN inside) doing some strange things > recently. If you were to send me a short program to compile and run on this > debian PowerPC (Mac G4) I can let you see the results. > > > > > PPS I made another test. In the big programm I have defined a function > > fermi2: > > > > ... > > > > and computed the following expression: > > > > aux22= 1.0d0-fermi2(enpct) > > > > The output was: > > > > on x86: > > tmp_fermi2( -2.20214243)= 1. > > aux22= 1.11022302E-16 !! should have been 0. !! > > > > This is an engineering department: 1.11022302E-16 == 0 ;-) Can you print out > the first expression more accuately? I bet it isn't really 1 (actually > 1.000000000000000111022302 ;-) Maybe there's no bug: it's just truncation > error? Maybe the maths unit on the alpha is better than the one on the ia32 > on > this test (no suprise there!)
Also, note that alpha is a 64bit plateform, and that as thus it could be that it handles some things better than ia32 or ppc. Not sure if this influences floating point values, i guess not. Friendly, Sven Luther
