[LLVMbugs] [Bug 12434] New: Polyhedron 2005 benchmark failures with -O2 -ftree-vectorize -fplugin-arg-dragonegg-enable-gcc-optzns
bugzilla-daemon at llvm.org
bugzilla-daemon at llvm.org
Sun Apr 1 12:27:37 PDT 2012
http://llvm.org/bugs/show_bug.cgi?id=12434
Bug #: 12434
Summary: Polyhedron 2005 benchmark failures with -O2
-ftree-vectorize
-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
Classification: Unclassified
A number of Polyhedron 2005 benchmarks are now failing when compiled with -O2
-ftree-vectorize -fplugin-arg-dragonegg-enable-gcc-optzns on
x86_64-apple-darwin11 at revision 153840. One example is doduc which when
compiled with...
de-gfortran46 -msse4 -O2 -ftree-vectorize
-fplugin-arg-dragonegg-enable-gcc-optzns -g doduc.f90 -o doduc
fails as...
% ./doduc
MAIN : FIN S00002
MAIN : FIN S00001
Illegal instruction
and backtraces in lldb as...
% lldb ./doduc
Current executable set to './doduc' (x86_64).
(lldb) r
Process 12839 launched: '/Users/howarth/Downloads/pb05/lin/source/doduc'
(x86_64)
MAIN : FIN S00002
MAIN : FIN S00001
Process 12839 stopped
* thread #1: tid = 0x1f03, 0x0000000100011d31 doduc`s00011_ + 2065 at
doduc.f90:2949, stop reason = EXC_BAD_INSTRUCTION (code=EXC_I386_INVOP,
subcode=0x0)
frame #0: 0x0000000100011d31 doduc`s00011_ + 2065 at doduc.f90:2949
2946 I6Ft = i
2947 J6Ft = j
2948 ENDIF
-> 2949 IF ( z8.LE.zmax ) THEN
2950 I8Ft = i
2951 J8Ft = j
2952 GOTO 200
(lldb) bt
* thread #1: tid = 0x1f03, 0x0000000100011d31 doduc`s00011_ + 2065 at
doduc.f90:2949, stop reason = EXC_BAD_INSTRUCTION (code=EXC_I386_INVOP,
subcode=0x0)
frame #0: 0x0000000100011d31 doduc`s00011_ + 2065 at doduc.f90:2949
frame #1: 0x0000000100019056 doduc`main [inlined] MAIN__ + 439 at
doduc.f90:169
frame #2: 0x0000000100018e9f doduc`main + 79 at doduc.f90:199
frame #3: 0x0000000100000d94 doduc`start + 52
(lldb)
using lldb from Xcode 4.3.2. The other Polyhedron 2005 benchmarks exhibiting
this failure include fatigue and test_fpu
--
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.
More information about the llvm-bugs
mailing list