We have a Fortran90 code which segfaults at a strange location when built with nvfortran and -O2 and higher. It runs fine with -O3 when built with other compilers (tried gfortran, ifort, ifx). Given that the crash only occurs with nvfortran and higher optimization I am wondering if this is not a compiler bug.
What would be the best way to bring it to the attention of the nvfortran developers?
Thanks.
Very likely.
What would be the best way to bring it to the attention of the nvfortran developers?
If you can post a reproducing example, I can file an issue report directly with the NVHPC compiler team. If the code can’t be posted publicly, you can direct message me, or I can email you and we can arrange a way you can share the code.
You can also submit a bug report yourself if you prefer.
Thanks Mat, I submitted a bug report.