Changes to HPC systems are listed below, optionally filtered by system.
The LAMMPS 14May16 known issue wherein parallel lammps spawned too many threads has been fixed on all clusters. No user action is required; if a user had applied the
OMP_NUM_THREADS workaround then it may be removed, but it will not cause probems if left in place. The corrected executables were made the defaults for module lammps/14may16 at these times:
module load caffe
Amber 16 has been installed on the OSC clusters; usage is via the module amber/16. For information on available executables and installation details see the software page for Amber or the output of the module help command, e.g.: module help amber/16. On August 15, 2016 Amber 16 will be made the default amber module.
LAMMPS stable version 14May16 has been installed on Oakley. Usage is via the module lammps/14May16. For information on installation details, such as, available packages, see the output of the module help command, e.g.: module help lammps/14May16
Comsol 5.2 has been installed on Ruby. To load, use
module load comsol/52
All ACLs set within the Home Directory filesystem (/nfs/##) were lost during the 6/7 downtime. This was caused by the migration to a new server that does not support the old POSIX ACLs.
Migrating the ACLs was not possible due to both the fact POSIX ACLs are not easily translatable to NFSv4 ACLs, and none of our tools supported such a migration.
Q-Chem 4.4 has been installed on Oakley. To load, use
module load qchem/4.4
In order to allow for a more equal distribution of resources amongst our community, the limit on the maximum cores in use by a user and by a group have both been temporarily reduced from 2040 cores to 1500 cores.
The max size of a parallel job has also been reduced from 170 nodes to 125 nodes.