Research computing team attends global supercomputing event

UAB's Research Computing team attended SC17, the premiere international supercomputing event, to learn how the latest advances in computing, storage and network technologies supporting global scientific initiatives can be applied to UAB research efforts and in support of Cheaha, UAB's high-performance computing cluster.

Dr. Puri Bangalore, interim director of research computing; John-Paul Robinson; Thomas Anthony; and Ravi Tripathi attended the November event in Denver and met with industry partners including Dell, IBM, Nvidia, and DDN to learn about the latest updates to the products that drive HPC at UAB. They learned about trends in HPC data center design and cooling technologies from vendors on the show floor and at dedicated tutorial sessions and met with academic peers from Stanford, Indiana University, Monash University, UTC, Eglin AFB, and Sandia National Laboratories to learn about how they are bringing high performance computing, cloud and integrated research services to their communities.

Bangalore organized the workshop on Exascale computing (ExaMPI).  Other members of the team attended workshops on directive-based GPU computing (WACCPD), medical imaging analysis and visualization (MIAV), HPC datacenter design, acquisition and commissioning, and use of Python in HPC (PyHPC).

Anthony was a featured speaker at the DDN booth and showcased the impact that UAB's multi-petabyte storage capacity is having on neuroimaging research at UAB. View his slides here.

About the SC Conference:

The Supercomputing Conference is an annual conference sponsored by the ACM Special Interest Group on High Performance Computing (SIGHPC) and the IEEE Computer Society. It is the premiere international conference for HPC and brings together vendors, researchers, and users of the worlds most advanced computing, storage and networking technologies for a week of meetings, workshops, tutorials, paper presentations, posters, and an expansive show floor.  SC17 was attended by more than 12,000 people.

The Supercomputing conference coincides with the fall release of the ranking for the Top500 super computers in the world.

Supercomputing features the annual construction of a state-of-the-art network named SCinet. SCinet is hub of the international networking community and is used to showcase the latest capabilities in high performance networking at the conference. The network provides production wired and wireless network services to conference attendees and vendors throughout the conference, making it an excellent demonstration of how cutting-edge research technologies can live side-by-side with production operations. SCinet is designed and built by over 200 volunteers from academia, government agencies, and industry from around the globe working together throughout the year. The network is custom built at the conference location in less than week, features over 50 miles of fiber optic cable.  This years' network supplied over 3Terabits network capacity, the equivalent of over thirty 100Gigabit university links like the one UAB installed last year.