November 20, 2012
Last week, the Mont-Blanc project announced it had tapped Samsung to supply the processors that would power an experimental high performance computing system.
Mont-Blanc is an EU-funded HPC architecture being developed at number of European supercomputing centers including Jülich Supercomputing Center and LRZ in Germany, CNRS and GENCI in France, CINECA in Italy, and Barcelona Supercomputing Center in Spain. The idea is to come up with a scalable architecture that will make it feasible to build exascale computers that run within a reasonable power envelope. To support the latter, the project is focusing on mobile/embedded computing technology, which has been key to delivering energy-efficient hardware.
The processor in question is Samsung's Exynos 5 Dual, an integrated ARM-GPU SoC built on its 32nm High-K Metal Gate (HKMG) process technology. It was originally designed for mobile consumer devices and is being used to power the Samsung Chromebook and the Google Nexus 10.
For the Mount-Blanc prototype, the Exynos chip will be housed on Samsung's Arndale development board, which is aimed at mobile computing and game developers. The board can be picked up for $249, which is fairly expensive for a mobile dev kit, but a pittance for most HPC set-ups, even in quantity.
Exynos 5, the processor, is made up of a dual-core 1.7 GHz ARM Cortex A15 CPU and a quad-core ARM Mali-T604 GPU. Although the chip is built to serve mobile computing apps and 3D gaming, according to Samsung, it can support 72 gigaflops at double precision. (Most mobile GPUs on the market today are limited to single precision.) It also supports up to 12.8 GB/second of memory bandwidth.
"The Exynos 5 Dual packs the most powerful ARM processors with a programmable GPU in a low-power mobile device that would normally be in someone's pocket and running on a battery. Its performance density, energy efficiency, and low market price make it an extraordinary building block for prototyping a new generation of HPC systems," said Mont-Blanc project coordinator Alex Ramirez, in the Nov. 13 press release.
Much of project effort will now shift toward developing a software stack and porting scientific applications for this integrated CPU-GPU architecture. Although, thus far, the work has focused on ARM-based solutions, AMD uses a similar heterogeneous architecture in its APU processors, in this case, based on x86 CPUs and AMD GPUs. NVIDIA, too, has a CPU-GPU solution in its Tegra line.
In fact, the Mount-Blanc project originally built a pre-prototype system based on a Tegra ARM-GPU SoC plus a discrete high-end NVIDIA GPU. Apparently, the researchers decided there was enough computing muscle in Samsung's integrated SoC to serve as an HPC testbed. NVIDIA is planning to integrate a 64-bit ARM CPU and a future GPU design within the next couple of years as part of its "Project Denver" architecture, so the GPU-maker's silicon may yet reappear in a future iteration of the Mont-Blanc work.
10/30/2013 | Cray, DDN, Mellanox, NetApp, ScaleMP, Supermicro, Xyratex | Creating data is easy… the challenge is getting it to the right place to make use of it. This paper discusses fresh solutions that can directly increase I/O efficiency, and the applications of these solutions to current, and new technology infrastructures.
10/01/2013 | IBM | A new trend is developing in the HPC space that is also affecting enterprise computing productivity with the arrival of “ultra-dense” hyper-scale servers.
Ken Claffey, SVP and General Manager at Xyratex, presents ClusterStor at the Vendor Showdown at ISC13 in Leipzig, Germany.
Join HPCwire Editor Nicole Hemsoth and Dr. David Bader from Georgia Tech as they take center stage on opening night at Atlanta's first Big Data Kick Off Week, filmed in front of a live audience. Nicole and David look at the evolution of HPC, today's big data challenges, discuss real world solutions, and reveal their predictions. Exactly what does the future holds for HPC?