The impact of the nanoscale on computing systems

Abstract

Abstract-Nanoscale technologies provide both challenges and opportunities. We show that the issues and potential solutions facing designers are technology independent and arise mainly from shrinking device sizes and an increase in the number of devices available. We explore how it is possible to use some of the devices that will be available to help ease design complexity as well as overcome process related challenges such as limited layout freedom, increased defect densities, timing constraints, and power dissipation.

Reliable computing with unreliable components: Using separable environments to stabilize long-term information storage

“…Intrinsic and extrinsic volatility is a problem likely to be encountered by many Unconventional Computing fabrics, especially at the nano/molecular scale [10]. Our results suggest a possible solution to this problem using unsupervised, bidirectional plasticity rules which automatically repair previously established decision boundaries.…”

“…the network made no attempt to account for errant synapses) and the degradation in performance persisted throughout the experiment. In the adaptive case, all synaptic weights, including those producing unreliable output, were continuously updated according to the AHAH rule (10). Remarkably, unsupervised bidirectional plasticity was able to restore performance on the XOR classification task until the decision surface roughly recovered its initial configuration.…”

“…XOR example. The decision surface through time (sampled uniformly) for a static network (top), which lacked plasticity, and for a modifiable network (bottom), with plasticity governed by the AHAH rule (10). At the time step indicated, faults were injected into first layer nodes (radial basis functions), causing the corresponding synapses to become unreliable.…”

“…Biological networks, for example, must overcome random perturbations introduced by the constant recycling of synaptic proteins [25]. Indeed, any physical implementation of a connectionist network, whether biological or engineered, will be subject to random fluctuations as the individual synaptic elements approach nano/molecular-scales [10]. How can the information distributed across the ensemble of synaptic weights be preserved for long periods in the face of both extrinsic and intrinsic sources of variability?…”

Reliability Analysis and Improvement in Nano Scale Design

“…At the circuit level, defect mapping techniques at per chip basis are proposed for post fabrication reconfiguration to identify defective devices and utilize spare defect free devices as functional units. At system level, redundancy based techniques such as N modular redundancy use multiple copies of the same application is executed and then using the majority voting the correct result is determined [4].…”