The Future of Fully Homomorphic Encryption System: from a Storage I/O Perspective

2025-11-07

AI Summary

Fully homomorphic encryption (FHE) enables computation on encrypted data, yet its practical deployment is severely constrained by storage I/O bottlenecks—an issue hitherto unaddressed in systematic studies. Method: This work presents the first storage-I/O–centric performance analysis of FHE systems, conducting fine-grained profiling on both ASIC and GPU platforms to quantify throughput and latency under varying I/O workloads. Contribution/Results: Experiments reveal that I/O overhead degrades ASIC performance by up to 357× and GPU performance by up to 22×. Based on these findings, we propose an I/O-aware hardware–software co-optimization paradigm, establishing I/O efficiency as a critical prerequisite for unlocking the full acceleration potential of FHE hardware. Our study provides novel architectural design principles and empirical evidence to guide future FHE system development.

Technology Category

• Machine Learning: Hardware-aware ML
• Search and Optimization: Evaluation and Analysis
• Humans and AI: Other Foundations of Human Computation & AI

Application Category

• Security and Privacy: Large-scale security measurements
• Systems and Infrastructure for Web, Mobile and WoT: Web performance, measurement, and characterization
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs

Abstract

Fully Homomorphic Encryption (FHE) allows computations to be performed on encrypted data, significantly enhancing user privacy. However, the I/O challenges associated with deploying FHE applications remains understudied. We analyze the impact of storage I/O on the performance of FHE applications and summarize key lessons from the status quo. Key results include that storage I/O can degrade the performance of ASICs by as much as 357$ imes$ and reduce GPUs performance by up to 22$ imes$.

Problem

• Analyzing storage I/O impact on Fully Homomorphic Encryption performance
• Studying I/O challenges in deploying encrypted computation applications
• Quantifying performance degradation in ASICs and GPUs from storage bottlenecks

Innovation

• Analyzes storage I/O impact on FHE performance
• Reveals ASIC performance degradation by 357 times
• Shows GPU performance reduction by up to 22 times