Despite the promise of extremely efficient matrix-vector multiplication (MVM) by ReRAM crossbar arrays (RCAs), maintaining high accuracy has been challenging due to nonidealities such as wire resistance (also known as IR drop) and I-V nonlinearity (i.e., voltage-dependent conductance). For system architects, a fast method to accurately predict the MVM output of an RCA under nonidealities is highly desirable. While IR drop alone without I-V nonlinearity can be efficiently predicted, the existence of I-V nonlinearity makes the problem much harder. In this paper we propose a novel algorithm based on iterative refinement, which can predict with high accuracy the outcome of an MVM operation on an RCA in the presence of both I-V nonlinearity and IR drop. Our experiments using binary RCAs of different sizes demonstrate that our proposed method is order-of-magnitude more accurate than previous methods in terms of RMS error. We also present case studies predicting hardware-realistic accuracy of binarized neural networks on RCAs as well as nonideality-aware retraining, demonstrating the efficacy of our method for early design space exploration of ReRAM-based accelerators.
|Original language||English (US)|
|Title of host publication||2022 IEEE 40th International Conference on Computer Design (ICCD)|
|State||Published - Dec 19 2022|
Bibliographical noteKAUST Repository Item: Exported on 2022-12-23
Acknowledged KAUST grant number(s): 4704 CRG2021
Acknowledgements: This work was supported by IITP grant (IITP-2021-0-02052, ITRC support program) and NRF grant (No. 2020R1A2C2015066) funded by MSIT of Korea, and by Free Innovative Research Fund of UNIST (1.170067.01). The EDA tool was supported by the IC Design Education Center (IDEC), Korea. This work has been funded by grant number 4704 CRG2021 from King Abdullah University of Science and Technology.