English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

As process technology continues to scale aggressively, circuit aging in aneuromorphic hardware due to negative bias temperature instability (NBTI) andtime-dependent dielectric breakdown (TDDB) is becoming a critical reliabilityissue and is expected to proliferate when using non-volatile memory (NVM) forsynaptic storage. This is because an NVM requires high voltage and current toaccess its synaptic weight, which further accelerates the circuit aging in aneuromorphic hardware. Current methods for qualifying reliability are overlyconservative, since they estimate circuit aging considering worst-caseoperating conditions and unnecessarily constrain performance. This paperproposes RENEU, a reliability-oriented approach to map machine learningapplications to neuromorphic hardware, with the aim of improving system-widereliability without compromising key performance metrics such as execution timeof these applications on the hardware. Fundamental to RENEU is a novelformulation of the aging of CMOS-based circuits in a neuromorphic hardwareconsidering different failure mechanisms. Using this formulation, RENEUdevelops a system-wide reliability model which can be used inside adesign-space exploration framework involving the mapping of neurons andsynapses to the hardware. To this end, RENEU uses an instance of Particle SwarmOptimization (PSO) to generate mappings that are Pareto-optimal in terms ofperformance and reliability. We evaluate RENEU using different machine learningapplications on a state-of-the-art neuromorphic hardware with NVM synapses. Ourresults demonstrate an average 38\% reduction in circuit aging, leading to anaverage 18% improvement in the lifetime of the hardware compared to currentpractices. RENEU only introduces a marginal performance overhead of 5% comparedto a performance-oriented state-of-the-art.

Improving Dependability of Neuromorphic Computing With Non-Volatile Memory