On Tue, 28 Feb 2006, Brian Barrett wrote:
> I asked Robert for some more information off the list and finally
> figured out what is going on. Intel's compilers implicitly add -
> lpthread if you give the -parallel option.
But this still doesn't explain my findings. 'mpif77 --showme' output
doesn't include any thread related option and I'm sure that no such
option was used in the Makefile. So the only way I can see it
happening is from the compiler itself, which I could check with a
simple hello-like program:
# ifort -o fint fint.f
# ldd fint
libm.so.6 => /lib/tls/libm.so.6 (0x00c13000)
libcxa.so.5 => /opt/intel_cc_80/lib/libcxa.so.5 (0x00a9d000)
libunwind.so.5 => /opt/intel_cc_80/lib/libunwind.so.5 (0x00df6000)
libpthread.so.0 => /lib/tls/libpthread.so.0 (0x004a2000)
libc.so.6 => /lib/tls/libc.so.6 (0x00111000)
/lib/ld-linux.so.2 => /lib/ld-linux.so.2 (0x00324000)
(libcxa and libunwind do not appear in the previous CHARMM ldd output
as the -shared-libcxa option of ifort is used to avoid dependency on
Intel shared libs).
So, I still don't understand where the difference comes from...
> I'm not sure what the -xN flag does
Exclusive optimization for P4 Northwood. The code will only run on
this type of CPUs.
--
Bogdan Costescu
IWR - Interdisziplinaeres Zentrum fuer Wissenschaftliches Rechnen
Universitaet Heidelberg, INF 368, D-69120 Heidelberg, GERMANY
Telephone: +49 6221 54 8869, Telefax: +49 6221 54 8868
E-mail: Bogdan.Costescu_at_[hidden]
|
