NTT-PIM: Row-Centric Architecture and Mapping for Efficient Number-Theoretic Transform on PIM
Authors:
- 9 July 2023
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Abstract
Recently DRAM-based PIMs (processing-in-memories) with unmodified cell arrays have demonstrated impressive performance for accelerating AI applications. However, due to the very restrictive hardware constraints, PIM remains an accelerator for simple functions only. In this paper we propose NTT-PIM, which is based on the same principles such as no modification of cell arrays and very restrictive area budget, but shows state-of-the-art performance for a very complex application such as NTT, thanks to features optimized for the application’s characteristics, such as in-place update and pipelining via multiple buffers. Our experimental results demonstrate that our NTT-PIM can outperform previous best PIM-based NTT accelerators in terms of runtime by 1.7 ∼ 17× while having negligible area and power overhead.Keywords
All Related Versions
Funding Information
- Samsung
- Samsung
This publication has 17 references indexed in Scilit:
- Homomorphic Encryption for Arithmetic of Approximate Numbers2017
- Architecting an Energy-Efficient DRAM System for GPUs2017
- Low-cost and area-efficient FPGA implementations of lattice-based cryptography2013
- CACTI-3DD: Architecture-level modeling for 3D die-stacked DRAM main memory2012
- Towards Efficient Arithmetic for Lattice-Based Cryptography on Reconfigurable Hardware2012
- On Ideal Lattices and Learning with Errors over Rings2010
- Vector radix fast Fourier transform2005
- Modular Multiplication Without Trial DivisionMathematics of Computation, 1985
- FFT algorithms for vector computersParallel Computing, 1984
- An Adaptation of the Fast Fourier Transform for Parallel ProcessingJournal of the ACM, 1968