Emerging Resistive Random Access Memory (RRAM) has shown the great potential of in-memory processing capability, and thus attracts considerable research interests in accelerating memory-intensive applications, such as neural networks. However, the accuracy of RRAM-based NN computing can degrade significantly, due to the intrinsic statistical variations of the resistance of RRAM cells. In this paper, we propose SIGHT, a SynergIstic alGorithm-arcHitecture fault-Tolerant framework, to holistically address this issue. Specifically, we consider three major types of faults for RRAM computing: non-linear resistance distribution, static variation, and dynamic variation. From the algorithm level, we propose a resistance-aware quantization to compel the neural network parameters to follow the exact non-linear resistance distribution as RRAM, and introduce an input regulation technique to compensate for RRAM variations. We also propose a selective weight refreshing scheme to address the dynamic variation issue that occurs at run-time. From the architecture level, we propose a general and low-cost architecture accordingly for supporting our fault-tolerant scheme. Our evaluation demonstrates almost no accuracy loss for our three fault-tolerant algorithms, and the proposed SIGHT architecture incurs performance overhead as little as 7.14%.
|Journal||IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems|
|Publication status||Accepted/In press - 2020|
All Science Journal Classification (ASJC) codes
- Computer Graphics and Computer-Aided Design
- Electrical and Electronic Engineering