[LLVMbugs] [Bug 10892] New: Polyhedron 2005 fatigue.f90 benchmark miscompiled with -msse4 -ffast-math -funroll-loops -O3 -fplugin-arg-dragonegg-enable-gcc-optzns

[bugzilla-daemon at llvm.org]

http://llvm.org/bugs/show_bug.cgi?id=10892

           Summary: Polyhedron 2005 fatigue.f90 benchmark miscompiled with
                    -msse4 -ffast-math -funroll-loops -O3
                    -fplugin-arg-dragonegg-enable-gcc-optzns
           Product: dragonegg
           Version: trunk
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Severity: normal
          Priority: P
         Component: New Bugs
        AssignedTo: baldrick at free.fr
        ReportedBy: howarth at nitro.med.uc.edu
                CC: llvmbugs at cs.uiuc.edu


At r139290, the Polyhedron 2005 fatigue.f90 benchmark is miscompiled with...

/sw/lib/gcc4.6/bin/gfortran -fplugin=/sw/lib/gcc4.6/lib/dragonegg.so -msse4
-ffast-math -funroll-loops -O3 -fplugin-arg-dragonegg-enable-gcc-optzns -g
fatigue.f90 -o fatigue

producing the incorrect run-time results of...

[MacPro:pb05/lin/source] howarth% ./fatigue


-------> Material failure in region    2 at r=     2.72730E+04 at cycle        
1


-------> Material failure in region    3 at r=     5.45461E+04 at cycle        
1


-------> Material failure in region    4 at r=     8.18191E+04 at cycle        
1


-------> Material failure in region    5 at r=     1.09092E+05 at cycle        
1


-------> Material failure in region    6 at r=     1.36365E+05 at cycle        
1


-------> Material failure in region    7 at r=     1.63638E+05 at cycle        
1


-------> Material failure in region    8 at r=     1.90911E+05 at cycle        
1

Simulation terminated.
     Spin cycle =                1
     Time (sec)       =     5.55556E-06
The wire has fractured, total damage =     2.40518E+09

================= Elasped CPU time  =     0.00000E+00

instead of the correct results when -fplugin-arg-dragonegg-enable-gcc-optzns
isn't used...


[MacPro:pb05/lin/source] howarth% ./fatigue

================= Spin cycle number =          1
================= Simulation time   =     2.77778E-04
================= Elasped CPU time  =     0.00000E+00



-------> Material failure in region   10 at r=     6.03250E-05 at cycle    
26102


-------> Material failure in region    9 at r=     5.39750E-05 at cycle    
37161


-------> Material failure in region    8 at r=     4.76250E-05 at cycle    
63210

Simulation terminated.
     Spin cycle =            63210
     Time (sec)       =     1.75581E+01
The wire has fractured, total damage =     5.34630E-01

================= Elasped CPU time  =     9.14900E+00

-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.