NDSU researchers, students and faculty can invest in condominiums in unexpected places by contacting computer experts instead of real estate agents. A new high performance computing resource called “Big Thunder”* provides additional options to scientific researchers and educators.
“These facilities will allow researchers access to additional state-of-the-art computing resources, where ‘big data’ analytics are efficiently coupled to high-performance modeling and simulation environments,” said Dr. Martin Ossowski, co-principal investigator on the Thunder Project and director of the NDSU Center for Computationally Assisted Science and Technology (CCAST), which houses Big Thunder.
In today’s competitive research funding environment, researchers and educators often use high-performance computers with significant power to conduct experiments and create models that may use millions of data points. As researchers’ computational needs at NDSU grow, what’s called a condominium model allows researchers and research groups to add their own hardware modules to Big Thunder for economies of scale.
Theoretical peak performance of Big Thunder nears 36 teraflops. One TFLOP equals 1012 floating point operations per second. The cluster’s file system is capable of > 4 Gbyte/second input/output throughput. The system can store ~4.5 PB of uncompressed data (>10 PB of compressed data. One PB = 1024 TB, One TB = 1012 bytes.)
“The rapid growth of computational power, along with important developments in computationally-driven science and engineering, will aid in major discoveries,” said Dr. Dinesh Katti, principal investigator for the Thunder project and interim chair of the NDSU Department of Civil and Environmental Engineering.
Big Thunder also provides large memory nodes for in-situ “big data” analysis, and a series of development nodes for convenient code development, optimization, and benchmarking.
On the data storage side, researchers and educators who need more Thunder than is included with a standard CCAST account can also participate in the system by providing their own media, such as disks and tapes. Individual researchers and educators can also add trays of disks, disk controllers, or library frames to Big Thunder at a lower price tag than a separate system with similar capabilities would cost. On the compute side, researchers can add their own combinations of CPUs, Intel Phis, or NVIDIA GPGPUs, with nearly exclusive control of associated queuing parameters.
Technical highlights of Big Thunder
- tiered General Parallel File System (GPFS) coupled seamlessly with the Hierarchical Storage Management (HSM) tape storage for transparent, policy based data movement
- Intel Ivy Bridge CPUs and Intel Phi co-processors
- FDR Infiniband interconnect
- serial attached SCSI (SAS) drives for increased Input/Output Operations Per Second (IOPS)
- file system optimized for broad spectrum of file types and sizes
Funding for Big Thunder at NDSU was provided by the NDSU Office of the Provost, the U.S. Department of Energy, and the National Science Foundation.
NDSU computational researchers contributed to the NSF proposal to receive competitive funding. Professor Dinesh Katti, civil engineering, serves as principal investigator. Co-principal investigators include: Anne Denton, Samee Khan, Martin Ossowski (CCAST director), and Wenfang Sun. NDSU faculty contributing to the proposal include Adnan Akyuz, Xeufeng Chu, Doğan Çömez, Sivaguru Jayaraman, Kalpana Katti, Svetlana Kilina, Ghodrat Karami, Andrei Kryjevski, Juan Li, Simone A. Ludwig, William Perrizo, Saeed Salem, Alexander Wagner, Yechun Wang, Changhui Yan, Mijia Yang, and Mariusz Ziejewski.
*Big Thunder, a derivative of Thundar, NDSU’s mascot, is used with permission of NDSU Athletics.