News
CMC’s high performance computer up and running
By Patrick Mann, CMC Director of IT
CMC-NCE and IPAC-CO2 maintain a high performance computer (HPC) which is generally free for use by members of the respective carbon management communities.
The high performance computer is a 128-node cluster with high-speed Infiniband interconnects and a software suite currently focused on large-scale modeling and simulation.
Our users are mostly graduate students who are running jobs for their research. For instance we have users modeling the performance of vapour extraction processes (Farid Amadloo) and novel enhanced oil recovery techniques (Harsh Joshi).
Flexibility key to system
We run the system relatively informally and can respond in a very flexible manner to user requests. In particular we entertain requests for special software packages for specific Carbon Management purposes. Currently we support the usual base software in a standard Linux distribution (CentOS) including the Gnu fortran and C++ compilers with the OpenMPI distributed programming libraries. However our users generally use various self-contained packages which we provide. Currently we have Fluent, CMG and MFIX.
As users may know we had a problem with our cluster in early May, and are pleased to announce that the system is back up and running. We ended up with a new installation using IBM’s xCAT cluster operating system which has given us considerably more flexibility in keeping the system updated. We can now for instance install most of the usual open-source scientific packages using standard update tools. This considerably enhances our ability to respond to user needs. In the process we added a separate fileserver, so we now have lots of disk space again. Hopefully users won’t fill it up too soon!
Looking for data sets
One new interest we have is in the provision of useful data sets. For instance we are investigating machine-readable versions of the Facility Greenhouse Gas data provided by NRCan. This kind of data can be very useful to many of our researchers, and machine-readable interfaces give users a change to build customized analysis and visualization applications. Are there any particular data sets that would be of value to your research and development? Give us a call!
We still have some compute cycles left, so if you would like to use the cluster please call the Carbon Commons IT Director Patrick Mann (patrick.mann@ipac-co2.com or 403-210 7058).
If you would like all the details see the High Performance Computing group on the Carbon Commons
