ccSEyJuT.out is attached generated after adding "-freport-bug"
to build/platforms/cuda/sharedTarget/CMakeFiles/OpenMMCUDA.dir/flags.make file
initial value:
CXX_FLAGS = -I/d12/shared/cuda/12.8.0_570.86.10/include -O2 -DNDEBUG -fPIC
-msse2 -DLEPTON_USE_JIT -DOPENMM_COMMON_BUILDING_SHARED_LIBRARY -std=gnu++11
changed into:
CXX_FLAGS = -I/d12/shared/cuda/12.8.0_570.86.10/include -O2 -DNDEBUG -fPIC
-msse2 -DLEPTON_USE_JIT -DOPENMM_COMMON_BUILDING_SHARED_LIBRARY -std=gnu++11
-freport-bug
```
during GIMPLE pass: vect
/jade/home/tru/build/openmm/openmm-8.2.0/platforms/common/src/CommonKernels.cpp:
In member function ‘void OpenMM::Common
CalcGayBerneForceKernel::sortAtoms()’:
/jade/home/tru/build/openmm/openmm-8.2.0/platforms/common/src/CommonKernels.cpp:5375:6:
internal compiler error: in vect
_get_vec_defs_for_operand, at tree-vect-stmts.cc:1509
5375 | void CommonCalcGayBerneForceKernel::sortAtoms() {
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~
0x1afaeb6 internal_error(char const*, ...)
???:0
0x69ec5a fancy_abort(char const*, int, char const*)
???:0
0x1abda20 vect_transform_stmt(vec_info*, _stmt_vec_info*,
gimple_stmt_iterator*, _slp_tree*, _slp_instance*)
???:0
0xfa2dbd vect_transform_loop(_loop_vec_info*, gimple*)
???:0
Please submit a full bug report, with preprocessed source.
Please include the complete backtrace with any bug report.
See <file:///usr/share/doc/gcc-12/README.Bugs> for instructions.
Preprocessed source stored into /tmp/ccSEyJuT.out file, please attach this to
your bugreport.
```
Other data points:
1) lowering the optimisation from -O3 to -O2 avoids the internal compiler error
2) using gcc-11/g++-11 (with -O3) instead og gcc-12/g++-12 does not cause any
issue.
Best regards
Tru
--
Tru Huynh (PhD) | mailto:t...@pasteur.fr | tel +33 1 45 68 87 37
https://research.pasteur.fr/en/team/structural-bioinformatics/
https://research.pasteur.fr/en/team/computational-structural-biology/
Institut Pasteur, 25-28 rue du Docteur Roux, 75724 Paris CEDEX 15 France
