Jeff Squyres wrote:
> Please run "file liblam_totalview.so.0.0.0.0".
Probably you meant file "liblam_totalview.so.0.0.0" - i don't have "file
liblam_totalview.so.0.0.0.0". Here's the output:
[ilashuk_at_master lam]$ file liblam_totalview.so.0.0.0
liblam_totalview.so.0.0.0: ELF 32-bit LSB shared object, Intel 80386,
version 1 (SYSV), not stripped
> Just curious -- do you only have one MPI implementation installed on
> your cluster (LAM/MPI 7.1.1)? If you have more than one, are you sure
> that you have your PATH and LD_LIBRARY_PATH setup properly such that
> they point to the 7.1.1 install?
I also have mpich2 and mpich 1.2.6 installed. lam and mpich2 are
installed in my home dir, mpich-1.2.6 is installed as root. I do not
have "LD_LIBRARY_PATH" in my environment. "which" tells me that mpirun,
mpicc, etc. come from lam.
> What was the exact command line that you used to configure LAM/MPI?
> (look in the first few lines of config.log in the LAM source directory)
./configure FFLAGS=-O0 CFLAGS=-O0 CXXFLAGS=-O0
--prefix=/home/guests/ilashuk/lam_install
Turning off optimization was just a last resort (maybe without much
sense) to make it work with hypre using "sysv" or "usysv" rpis.
Totalview problems emerged as well when configured just with
"./configure --prefix=/home/guests/ilashuk/lam_install", to my recollection.
|
