Abstract
Stochastic computing (SC) is a special type of digital compute strategy where values are represented by the probability of 1 and 0 in stochastic bit streams, which leads to superior hardware simplicity and error-tolerance. In this letter, we propose and demonstrate SC with GeSe-based Ovonic Threshold Switching (OTS) selector devices by exploiting their probabilistic switching behavior. The stochastic bit streams generated by OTS are demonstrated with good computation accuracy in both multiplication operation and image processing circuit. Moreover, the bit distribution has been statistically studied and linked to the collective defect de/localization behavior in the chalcogenide material. Weibull distribution of the delay time supports the origin of such probabilistic switching, facilitates further optimization of the operation condition, and lays the foundation for device modelling and circuit design. Considering its other advantages such as simple structure, fast speed, and volatile nature, OTS is a promising material for implementing SC in a wide range of novel applications, such as image processors, neural networks, control systems and reliability analysis.
Accepted Version (
Free)
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call