> > > > Are there newer versions of the Gnu compiler for this system?
Yes: -------------------------------------------------------------------------------------------------- /opt/modulefiles -------------------------------------------------------------------------------------------------- gcc/4.8.1 gcc/4.8.2 gcc/4.9.3 gcc/5.3.0 gcc/6.1.0 gcc/6.2.0 gcc/6.3.0(default) gcc/7.1.0 gcc/7.2.0 gcc/7.3.0 > Are there any other compilers on the system that would likely be less > buggy? IBM compilers? If this simple code generates a gross error with > optimization who's to say how many more subtle bugs may be induced in the > library by the buggy optimizer (there may be none but IMHO probability says > there will be others). > Let me ask them what they recommend to use with cuda codes. > > Is there any chance that valgrind runs on this machine; you could run > the optimized version through it and see what it says. > > Valgrind works but tons of output and I could not see anything interesting in there. And, this test does work with 1 processor! I think this is only a problem when GMRES is uses as the eigen estimator in Cheby. GMRES solvers work and Cheby works with -mg_levels_esteig_ksp_type cg.