High-Performance Deterministic Stochastic Computing Using Residue Number System

Editor’s notes: This article discusses how to reduce the latency of stochastic computations. The authors represent an integer number as a set of remainders with respect to a set of relatively prime moduli. Operations such as multiplication, implemented using a deterministic version of stochastic computing, work directly on the remainders, thus yielding a partitioning of the original computation and a significant decrease in the number of clock cycles required. —Vincent T. Lee, Facebook Reality Labs Research

[1]  Yuke Wang.  Residue-to-binary converters based on new Chinese remainder theorems , 2000 .

[2]  H. Garner.  The residue number system , 1959, IRE-AIEE-ACM '59 (Western).

[3]  Ricardo Chaves,et al.  Method to Design General RNS Reverse Converters for Extended Moduli Sets , 2013, IEEE Transactions on Circuits and Systems II: Express Briefs.

[4]  Chip-Hong Chang,et al.  Residue Number Systems: A New Paradigm to Datapath Optimization for Low-Power and High-Performance Digital Signal Processing Applications , 2015, IEEE Circuits and Systems Magazine.

[5]  Brian R. Gaines,et al.  Stochastic Computing Systems , 1969 .

[6]  Tajana Simunic,et al.  RNSnet: In-Memory Neural Network Acceleration Using Residue Number System , 2018, 2018 IEEE International Conference on Rebooting Computing (ICRC).

[7]  M. Hassan Najafi,et al.  Late Breaking Results: LDFSM: A Low-Cost Bit-Stream Generator for Low-Discrepancy Stochastic Computing , 2020, 2020 57th ACM/IEEE Design Automation Conference (DAC).

[8]  David J. Lilja,et al.  Performing Stochastic Computation Deterministically , 2019, IEEE Transactions on Very Large Scale Integration (VLSI) Systems.

[9]  John P. Hayes,et al.  The Promise and Challenge of Stochastic Computing , 2018, IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems.

[10]  Marc D. Riedel,et al.  A deterministic approach to stochastic computation , 2016, 2016 IEEE/ACM International Conference on Computer-Aided Design (ICCAD).

[11]  Mehdi Kamal,et al.  Res-DNN: A Residue Number System-Based DNN Accelerator Unit , 2020, IEEE Transactions on Circuits and Systems I: Regular Papers.