November 06, 2013
Today's supercomputers are facing multiple challenges relating to the very processor technology that is these machines' lifeblood. As Moore's Law peters out, researchers are looking to alternatives to traditional silicon-based chips. Continuing the kind of 1000X improvements the community has grown accustomed to likely hinges on a breakthrough semiconductor material.
Materials scientists at the Harvard School of Engineering and Applied Sciences (SEAS) are looking to biology for inspiration, specifically to the human brain. Consider for a moment that the human brain, while interpreting the real world in real-time, uses roughly the same amount of power as a 20-watt light bulb.
The Harvard-based team has developed a transistor that behaves like a synapse. As in a real brain, the synaptic transistor actually learns while it computes. The advance could lead to a new type of artificial intelligence that is embedded in a computer's architecture.
With 86 billion neurons, connected by synapses, human brains are essentially highly-efficient, highly parallel computers. In addition to acting as logic circuits, the synapses continuously adapt to stimuli, strengthening some connections and weakening others. Scientists refer to this as process learning, and it's orders more efficient and elegant than the machine learning employed by Siri and Blue Gene.
Using the brain as a template – scientists are designing a system that incorporates millions of tiny synaptic transistors and neuron terminals. In doing so, they hope to bring parallel computing into a new era of ultra-efficient high performance.
"The transistor we've demonstrated is really an analog to the synapse in our brains," says co-lead author Jian Shi, a postdoctoral fellow at SEAS. "Each time a neuron initiates an action and another neuron reacts, the synapse between them increases the strength of its connection. And the faster the neurons spike each time, the stronger the synaptic connection. Essentially, it memorizes the action between the neurons."
The research team created the synaptic transistor with samarium nickelate, a correlated electron system that can undergo an insulator-metal transition. At a certain temperature, or when exposed to an external field, the conductance of the material suddenly changes, setting the stage for a highly energy-efficient switching mechanism.
"Structurally, the device consists of the nickelate semiconductor sandwiched between two platinum electrodes and adjacent to a small pocket of ionic liquid. An external circuit multiplexer converts the time delay into a magnitude of voltage which it applies to the ionic liquid, creating an electric field that either drives ions into the nickelate or removes them. The entire device, just a few hundred microns long, is embedded in a silicon chip," reads an official announcement.
The synaptic transistor offers several advantages, according to the researchers. Besides being inherently energy-efficient, it is not restricted to binary operations; it employs non-volatile memory; and it's compatible with existing silicon-based systems.
The Harvard scientists documented their work in Nature Communications paper, "A correlated nickelate synaptic transistor."
"Inspired by biological neural systems, neuromorphic devices may open up new computing paradigms to explore cognition, learning and limits of parallel computation,” they write.
This is Your Brain on GPUs (Datanami)
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?