FlexMem: High-Parallel Near-Memory Architecture for Flexible Dataflow in Fully Homomorphic Encryption

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Shi, Shangyi, Han, Husheng, Mu, Jianan, Zheng, Xinyao, Liang, Ling, Lu, Hang, Du, Zidong, Li, Xiaowei, Hu, Xing, Guo, Qi
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866913766639140864
author Shi, Shangyi
Han, Husheng
Mu, Jianan
Zheng, Xinyao
Liang, Ling
Lu, Hang
Du, Zidong
Li, Xiaowei
Hu, Xing
Guo, Qi
author_facet Shi, Shangyi
Han, Husheng
Mu, Jianan
Zheng, Xinyao
Liang, Ling
Lu, Hang
Du, Zidong
Li, Xiaowei
Hu, Xing
Guo, Qi
contents Fully Homomorphic Encryption (FHE) imposes substantial memory bandwidth demands, presenting significant challenges for efficient hardware acceleration. Near-memory Processing (NMP) has emerged as a promising architectural solution to alleviate the memory bottleneck. However, the irregular memory access patterns and flexible dataflows inherent to FHE limit the effectiveness of existing NMP accelerators, which fail to fully utilize the available near-memory bandwidth. In this work, we propose FlexMem, a near-memory accelerator featuring high-parallel computational units with varying memory access strides and interconnect topologies to effectively handle irregular memory access patterns. Furthermore, we design polynomial and ciphertext-level dataflows to efficiently utilize near-memory bandwidth under varying degrees of polynomial parallelism and enhance parallel performance. Experimental results demonstrate that FlexMem achieves 1.12 times of performance improvement over state-of-the-art near-memory architectures, with 95.7% of near-memory bandwidth utilization.
format Preprint
id arxiv_https___arxiv_org_abs_2503_23496
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle FlexMem: High-Parallel Near-Memory Architecture for Flexible Dataflow in Fully Homomorphic Encryption
Shi, Shangyi
Han, Husheng
Mu, Jianan
Zheng, Xinyao
Liang, Ling
Lu, Hang
Du, Zidong
Li, Xiaowei
Hu, Xing
Guo, Qi
Hardware Architecture
Fully Homomorphic Encryption (FHE) imposes substantial memory bandwidth demands, presenting significant challenges for efficient hardware acceleration. Near-memory Processing (NMP) has emerged as a promising architectural solution to alleviate the memory bottleneck. However, the irregular memory access patterns and flexible dataflows inherent to FHE limit the effectiveness of existing NMP accelerators, which fail to fully utilize the available near-memory bandwidth. In this work, we propose FlexMem, a near-memory accelerator featuring high-parallel computational units with varying memory access strides and interconnect topologies to effectively handle irregular memory access patterns. Furthermore, we design polynomial and ciphertext-level dataflows to efficiently utilize near-memory bandwidth under varying degrees of polynomial parallelism and enhance parallel performance. Experimental results demonstrate that FlexMem achieves 1.12 times of performance improvement over state-of-the-art near-memory architectures, with 95.7% of near-memory bandwidth utilization.
title FlexMem: High-Parallel Near-Memory Architecture for Flexible Dataflow in Fully Homomorphic Encryption
topic Hardware Architecture
url https://arxiv.org/abs/2503.23496