The US Department of Energy’s Oak Ridge National Laboratory [ORNL] announced the start-up of a new supercomputer, called Titan, today. The new machine, which is likely to displace the Sequoia computer at Lawrence Livermore National Laboratory as the world’s fastest supercomputer, has been in the works for two years.
Performance of these systems is ranked based on their speed in floating-point operations per second (FLOP/s), measured on the LINPACK benchmark, which involves the solution of a dense system of linear equations. The Sequoia, ranked fastest in the world in June of this year, achieved over 16 petaflops (1.6 × 1016 flops); the new Titan system is rated at 27 petaflops (2.7 × 1016 flops).
The Titan system uses a hybrid architecture that includes both conventional 16-core Opteron 6274 CPUs, and NVIDIA Tesla K-20 GPUs. It has a total of 18,688 compute nodes, each containing a GPU and a CPU, for a total of 299.008 CPU cores. The system also has more than 700 terabytes ( 7 × 1014 bytes) of memory. (Apparently 640KB is no longer enough for anyone.) The hybrid architecture results in better energy efficiency; Titan gets about ten times the performance of its predecessor at ORNL, the Jaguar, at less than 30% more electricity consumption. It is, however, rather large, requiring 4,352 square feet of floor space.
James Hack, Director of ORNL’s National Center for Computational Sciences, said “Titan will allow scientists to simulate physical systems more realistically and in far greater detail. The improvements in simulation fidelity will accelerate progress in a wide range of research areas such as alternative energy and energy efficiency, the identification and development of novel and useful materials and the opportunity for more advanced climate projections.”
The ORNL press release is here.