NVIDIA has just announced the pioneering efforts of the Barcelona Supercomputing Center (BSC) to build a new hybrid supercomputer that uses energy efficient, low-power NVIDIA Tegra ARM CPUs, together with high-performance Tesla GPUs. This prototype system is believed to be the world's first large-scale implementation of an ARM-based CPU/GPU supercomputing combination.
With this prototype system, the center aims to achieve a near term goal of improving energy efficiency by two to five times, compared with today’s most efficient system. Their ultimate goal is to reach exascale computing levels using 15 to 30 times less power than current supercomputer architectures. This massive undertaking has been named the EU Mont-Blanc Project and its objective is to showcase the energy efficiency of ARM-based processors without the need to compromise on computing power when paired with appropriate GPUs.
According to NVIDIA's director of Tesla marketing, Sumit Gupta who said that the use of graphics processors is nascent in the supercomputing space, their level of adoption is growing and a natural ecosystem of software applications is evolving to support their adoption.
As an attestation to their adoption for supercomputing requirements, another supercomputing project which is in progress is the Titan, a Cray XK6 supercomputer using Tesla GPUs that will have the potential to deliver over 20 petaflops of peak performance. Titan is also twice as fast and three times more energy efficient than today’s fastest supercomputer, the K Computer, located in Japan. The Titan project is an undertaking by the Oak Ridge National Laboratory (ORNL), an open science computing facility for the U.S. Department of Energy.
On its own initiative, NVIDIA is planning to develop a hardware and software development kit which will feature a quad-core NVIDIA Tegra 3 ARM CPU accelerated by a discrete NVIDIA GPU. This kit is supported by NVIDIA's CUDA programming toolkit and is expected to be available in the first half of 2012.
Source: NVIDIA, EE Times