Hi Derek,
try the -no-multibyte-chars flag when you build mpi compilers with icc.
Then the flag will be used also when you compile petsc with mpicc.
Maybe using the flag for petsc compilation works without recompiling mpi.
Lorenzo
On Feb 23, 2012, at 3:57 PM, Derek Gaston wrote:
> Apparentl
Apparently PETSc 3.2-p5 works fine weird.
Derek
On Thu, Feb 23, 2012 at 7:34 AM, Paul T. Bauman wrote:
> I don't have access to 12.1 ATM (only 12.0), but someone I work with
> complained to me the other day that a micro version change in 12.1 caused
> regressions in their code. So if it's t
I don't have access to 12.1 ATM (only 12.0), but someone I work with
complained to me the other day that a micro version change in 12.1 caused
regressions in their code. So if it's the latest and "greatest" version of
Intel, I might be wary.
On Thu, Feb 23, 2012 at 8:29 AM, Derek Gaston wrote:
>
Anyone else out there seen this lately:
Compiling C++ (in optimized mode) src/numerics/petsc_vector.C...
FATAL ERROR : Compiler Internal Error
compilation aborted for src/numerics/petsc_vector.C (code 2)
make: *** [src/numerics/petsc_vector.x86_64-unknown-linux-gnu.opt.o] Error 2
This is w
