Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

CAT: A GPU-Accelerated FHE Framework with Its Application to High-Precision Private Dataset Query

Qirui Li, Rui Zong

TL;DR

CAT introduces a GPU-accelerated, three-layer FHE framework that unifies core math, pre-computed elements, and API-level operators under two GPU pools to achieve large performance gains. It implements CKKS, BFV, and BGV, delivering up to 2173× speedups over CPU implementations and enabling CKKS-based PDQ queries that complete on 10^3 rows within 1 second using 2–5 GB of memory. A novel multiplicative-masking two-party protocol enables high-precision, bootstrapping-free encrypted "search-and-average" queries, expanding practical PDQ feasibility. The work demonstrates strong real-world potential for integrating state-of-the-art FHE into systems, provides open-source tooling, and outlines future enhancements in bootstrapping, scheme switching, and expanded query operators.

Abstract

We introduce an open-source GPU-accelerated fully homomorphic encryption (FHE) framework CAT, which surpasses existing solutions in functionality and efficiency. \emph{CAT} features a three-layer architecture: a foundation of core math, a bridge of pre-computed elements and combined operations, and an API-accessible layer of FHE operators. It utilizes techniques such as parallel executed operations, well-defined layout patterns of cipher data, kernel fusion/segmentation, and dual GPU pools to enhance the overall execution efficiency. In addition, a memory management mechanism ensures server-side suitability and prevents data leakage. Based on our framework, we implement three widely used FHE schemes: CKKS, BFV, and BGV. The results show that our implementation on Nvidia 4090 can achieve up to 2173$\times$ speedup over CPU implementation and 1.25$\times$ over state-of-the-art GPU acceleration work for specific operations. What's more, we offer a scenario validation with CKKS-based Privacy Database Queries, achieving a 33$\times$ speedup over its CPU counterpart. All query tasks can handle datasets up to $10^3$ rows on a single GPU within 1 second, using 2-5 GB storage. Our implementation has undergone extensive stability testing and can be easily deployed on commercial GPUs. We hope that our work will significantly advance the integration of state-of-the-art FHE algorithms into diverse real-world systems by providing a robust, industry-ready, and open-source tool.

CAT: A GPU-Accelerated FHE Framework with Its Application to High-Precision Private Dataset Query

TL;DR

CAT introduces a GPU-accelerated, three-layer FHE framework that unifies core math, pre-computed elements, and API-level operators under two GPU pools to achieve large performance gains. It implements CKKS, BFV, and BGV, delivering up to 2173× speedups over CPU implementations and enabling CKKS-based PDQ queries that complete on 10^3 rows within 1 second using 2–5 GB of memory. A novel multiplicative-masking two-party protocol enables high-precision, bootstrapping-free encrypted "search-and-average" queries, expanding practical PDQ feasibility. The work demonstrates strong real-world potential for integrating state-of-the-art FHE into systems, provides open-source tooling, and outlines future enhancements in bootstrapping, scheme switching, and expanded query operators.

Abstract

We introduce an open-source GPU-accelerated fully homomorphic encryption (FHE) framework CAT, which surpasses existing solutions in functionality and efficiency. \emph{CAT} features a three-layer architecture: a foundation of core math, a bridge of pre-computed elements and combined operations, and an API-accessible layer of FHE operators. It utilizes techniques such as parallel executed operations, well-defined layout patterns of cipher data, kernel fusion/segmentation, and dual GPU pools to enhance the overall execution efficiency. In addition, a memory management mechanism ensures server-side suitability and prevents data leakage. Based on our framework, we implement three widely used FHE schemes: CKKS, BFV, and BGV. The results show that our implementation on Nvidia 4090 can achieve up to 2173 speedup over CPU implementation and 1.25 over state-of-the-art GPU acceleration work for specific operations. What's more, we offer a scenario validation with CKKS-based Privacy Database Queries, achieving a 33 speedup over its CPU counterpart. All query tasks can handle datasets up to rows on a single GPU within 1 second, using 2-5 GB storage. Our implementation has undergone extensive stability testing and can be easily deployed on commercial GPUs. We hope that our work will significantly advance the integration of state-of-the-art FHE algorithms into diverse real-world systems by providing a robust, industry-ready, and open-source tool.

Paper Structure

This paper contains 26 sections, 1 equation, 12 figures, 7 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Background and Motivation
  3. Notations
  4. RNS Representation
  5. FHE Hardware Accelration
  6. FHE-based Private Database Query
  7. Architecture Design Overview
  8. A three leveled acceleration framework
  9. Convert Operator to SIMD Paradigm
  10. Multi Level Memory Usage on GPU
  11. Kernel Fusion and Segmentation
  12. Date layout and Indexing
  13. Utilizing Multi stream
  14. Design of GPU-memory pool for CAT
  15. Architecture of Memory pool
  16. ...and 11 more sections

Figures (12)

  • Figure 1: Workflow for a PDQ Query
  • Figure 2: Overall architecture of the GPU-accelerated FHE framework
  • Figure 3: Structure for storing underlying data
  • Figure 4: Architecture of Memory pool
  • Figure 5: Time consumption of each stages without any optimization
  • ...and 7 more figures