ARM Holdings plc (ADR) (NASDAQ:ARMH) (LON:ARM) announced on Monday, June 23rd, that its first 64-bit ARM-based systems designed for supercomputers are expected to ship later this year. The high performance systems will run the Applied Micro X-Gene system-on-a-chip, with 64-bit ARMv8-based processor cores and NVIDIA Corporation (NASDAQ:NVDA) Tesla K20 GPU-based accelerators.
ARM expanding out of mobile into datacenters
The vast majority of ARM chips today are used to power phones and tablets, but ARM Holdings plc (ADR) (NASDAQ:ARMH) (LON:ARM) and its partners have been eying an expansion into supercomputers and datacenters for a while now, and we will see the launch of a number of new SoCs based on ARM’s v8 architecture over the next few quarters.
According to industry sources, Dell and HP are planning to release servers based on Applied Micro Circuits Corporation (NASDAQ:AMCC)’s X-Gene processor, and before the end of the year AMD is expected to launch its ARMv8-based Opteron A1100 processor, a multicore SoC expandable up to 128GB of DRAM.
ARM Holdings plc (ADR) (NASDAQ:ARMH) (LON:ARM) developed the 64-bit v8 architecture for the server market, and forthcoming ARM-based SoCs are focused around support for enterprise-class features, including error-correcting code memory.
The new ARM-based servers are designed to handle hyperscale computing workloads, lighter-weight tasks to be carried out on a huge scale, such as managing web content for a large website. The demand for hardware capable of hyperscale workloads is growing, with tech titans such as Google, Facebook and cloud services providers lining up to buy them. According to ZDNet sources, Google Inc (NASDAQ:GOOG) (NASDAQ:GOOGL) and Facebook Inc (NASDAQ:FB) are considering moving ARM-based silicon server chip systems with architecture customized to their needs.
New HPC systems
The ARM-based high performance computing systems to be released in the coming months are produced by Cirrascale Corp, E4 Computer Engineering and the Eurotech Group.
- High-density two-in-one 1U server with two Tesla K20 GPU accelerators, with network and storage offload engines, including integrated Ethernet. This server is optimized for private cloud, public cloud, HPC and enterprise applications. The Cirrascale system is available to order and will begin shipping in the third quarter of 2014.
- Low-power 3U, dual-motherboard server with two Tesla K20 GPU accelerators, designed from the ground up for seismic, signal and image processing, video analytics, track analysis, web applications and MapReduce processing.
- Ultra-high density, energy efficient and modular Aurora HPC server device, utilizing proprietary Brick Technology with direct hot liquid cooling.
These new hyperscale systems will run the NVIDIA CUDA 6.5 parallel programming platform. All are currently designed to be development systems, but production versions are expected to be released by the end of the year.