This is the mail archive of the
fortran@gcc.gnu.org
mailing list for the GNU Fortran project.
Regression in some polyhedron tests
- From: dominiq at lps dot ens dot fr (Dominique Dhumieres)
- To: fortran at gcc dot gnu dot org
- Date: Sat, 24 Feb 2007 20:52:33 +0100 (CET)
- Subject: Regression in some polyhedron tests
With powerpc-apple-darwin7.9.0, version 4.3.0 20070223,
the polyhedron test 'fatigue.f90' gives
...
Simulation terminated.
This sample will never fracture.
The imposed stress is below the fatigue limit.
See the output file for more information.
with optimisation -O1 and above. Also the output of induct
is not correct with optimization -O1 and above:
...
Minimum wand/quad abs rel mutual inductance = 0.00000000000000
Maximum wand/quad abs rel mutual inductance = 5.953794284446593E-002
Minimum resq coil/quad abs rel mutual inductance = 0.00000000000000
Maximum resq coil/quad abs rel mutual inductance = 9.639952502420604E-002
...
Number of wand/quad mutuals computed = 360000
Number of wand/quad mutuals actually used = 241200
without optimization,
...
Minimum wand/quad abs rel mutual inductance = 1.000000000000000E+020
Maximum wand/quad abs rel mutual inductance = -1.000000000000000E+020
Minimum resq coil/quad abs rel mutual inductance = 0.00000000000000
Maximum resq coil/quad abs rel mutual inductance = 2.765073637794563E-002
...
Number of wand/quad mutuals computed = 360000
Number of wand/quad mutuals actually used = 0
...
with -O1,
...
Minimum wand/quad abs rel mutual inductance = 2.032717691019597E-315
Maximum wand/quad abs rel mutual inductance = 9.943303690949241E-002
Minimum resq coil/quad abs rel mutual inductance = 0.00000000000000
Maximum resq coil/quad abs rel mutual inductance = 2.765073637794563E-002
...
Number of wand/quad mutuals computed = 360000
Number of wand/quad mutuals actually used = 121200
...
with -O2 and -O3,
...
Minimum wand/quad abs rel mutual inductance = 1.047631426701815E-314
Maximum wand/quad abs rel mutual inductance = 0.495335719303436
Minimum resq coil/quad abs rel mutual inductance = 0.00000000000000
Maximum resq coil/quad abs rel mutual inductance = 2.765073637794563E-002
...
Number of wand/quad mutuals computed = 360000
Number of wand/quad mutuals actually used = 241200
with -O3 -funroll-loops,
...
Minimum wand/quad abs rel mutual inductance = 1.000000000000000E+020
Maximum wand/quad abs rel mutual inductance = NaN
Minimum resq coil/quad abs rel mutual inductance = 1.592081921152750E-236
Maximum resq coil/quad abs rel mutual inductance = 2.765073637794560E-002
...
Number of wand/quad mutuals computed = 360000
Number of wand/quad mutuals actually used = 360000
with -O3 -ffast-math -funroll-loops.
Does anyone has an idea of where the problems are coming from? The last week
snapshot worked without problem.
Can someone repeats the tests under OSX 10.4 for a version between
20070217 and 20070223?
TIA
Dominique