F1: A Fast and Programmable Accelerator for Fully Homomorphic Encryption (Extended Version)

TL;DR

F1 is a programmable hardware accelerator designed specifically for fully homomorphic encryption, achieving unprecedented speedups over software implementations and enabling practical real-time privacy-preserving computations.

Contribution

F1 introduces the first fully programmable FHE accelerator with specialized functional units and a data movement optimized architecture, significantly improving FHE computation speed.

Findings

F1 outperforms software by up to 17000x in FHE tasks.

F1 enables real-time private deep learning applications.

F1 is the first to accelerate complete FHE programs.

Abstract

Fully Homomorphic Encryption (FHE) allows computing on encrypted data, enabling secure offloading of computation to untrusted serves. Though it provides ideal security, FHE is expensive when executed in software, 4 to 5 orders of magnitude slower than computing on unencrypted data. These overheads are a major barrier to FHE's widespread adoption. We present F1, the first FHE accelerator that is programmable, i.e., capable of executing full FHE programs. F1 builds on an in-depth architectural analysis of the characteristics of FHE computations that reveals acceleration opportunities. F1 is a wide-vector processor with novel functional units deeply specialized to FHE primitives, such as modular arithmetic, number-theoretic transforms, and structured permutations. This organization provides so much compute throughput that data movement becomes the bottleneck. Thus, F1 is primarily designed to minimize data movement. The F1 hardware provides an explicitly managed memory hierarchy and mechanisms to decouple data movement from execution. A novel compiler leverages these mechanisms to maximize reuse and schedule off-chip and on-chip data movement. We evaluate F1 using cycle-accurate simulations and RTL synthesis. F1 is the first system to accelerate complete FHE programs and outperforms state-of-the-art software implementations by gmean 5400x and by up to 17000x. These speedups counter most of FHE's overheads and enable new applications, like real-time private deep learning in the cloud.