Skip to content

Phantom: A CUDA-Accelerated Word-Wise Homomorphic Encryption Library

Authors: H. Yang; S. Shen; W. Dai; L. Zhou; Z. Liu; Y. Zhao
Year: 2024
Venue: IEEE TDSC 2024

type: library collection: catalogue, sok-references status: auto

Links: ePrint 2023/49 · link

Summary

Auto-extracted summary (unreviewed)

Phantom is a CUDA library implementing the three word-wise HE schemes BGV, BFV and CKKS in one unified framework, with an improved hybrid key-switching technique, several kernel-fusion strategies, optimised memory access patterns and a memory pool, plus management of precomputation and RNS bases. The authors compare against state-of-the-art GPU implementations and SEAL and provide a cross-scheme benchmark on the same platform intended as a reference for scheme selection.

Abstract

Homomorphic encryption (HE) is a promising technique for privacy-preserving computations, especially the word-wise HE schemes that allow batching. However, the high computational overhead hinders the deployment of HE in real-word applications. GPUs are often used to accelerate execution, but a comprehensive performance comparison of different schemes on the same platform is still missing. In this work, we fill this gap by implementing three word-wise HE schemes BGV, BFV, and CKKS on GPU, with both theoretical and engineering optimizations. We enhance the hybrid key-switching technique, significantly reducing the computational and memory overhead. We explore several kernel fusing strategies to reuse data, resulting in reduced memory access and IO latency, and enhancing the overall performance. By comparing with the state-of-the-art works, we demonstrate the effectiveness of our implementation. Meanwhile, we introduce a unified framework that finely integrates our implementation of the three schemes, covering almost all scheme functions and homomorphic operations. We optimize the management of pre-computation, RNS bases, and memory in the framework, to provide efficient and lowlatency data access and transfer. Based on this framework, we provide a thorough benchmark of the three schemes, which can serve as a reference for scheme selection and implementation in constructing privacy-preserving applications. Our library is available for access at https://github.com/encryptorion-lab/phantom-fhe. It is released under the GPLv3 license.
Source: pdf

Extracted details

Field Value Source Confidence
contribution GPU implementations of BGV, BFV and CKKS llm 0.80
contribution enhanced hybrid key-switching with lower computational and memory overhead llm 0.80
contribution intra- and inter-kernel fusion strategies llm 0.80
contribution unified framework with memory pool, precomputation and RNS base management llm 0.80
contribution thorough same-platform benchmark of the three schemes llm 0.80
schemes BGV llm 0.80
schemes BFV llm 0.80
schemes CKKS llm 0.80
libraries Phantom llm 0.80
libraries CUDA llm 0.80
libraries SEAL llm 0.80
tasks GPU acceleration of HE primitives llm 0.80
tasks benchmarking llm 0.80
hardware NVIDIA GPU (CUDA) llm 0.80
security level 128-bit llm 0.80
interaction non-interactive FHE llm 0.80
code url https://github.com/encryptorion-lab/phantom-fhe (URL found in the full text) regex 0.50
code url https://github.com/KyoohyungHan/FullRNS-HEAAN (URL found in the full text) regex 0.50
code url https://github.com/snucrypto/HEAAN (URL found in the full text) regex 0.50
code url https://github.com/encryptorion-lab/phantom-fhe llm 0.80
llm relevance low llm 0.80

System card: Phantom

Group 7. FHE libraries, GPU back-ends and compilers
Family CKKS, BFV, BGV
Arithmetic CUDA C++
Interaction GPU library
Model scale Used by NEXUS, MOAI, EUSTON, STIP, ENSI, CACHEMIR
Status Widely used
Relation to the SoK Open GPU CKKS library used by most SoK GPU frameworks.
Note GPL-3.0. Bootstrapping added by downstream frameworks.
Code https://github.com/encryptorion-lab/phantom-fhe
Systems page Phantom

SoK reference [88]

H. Yang, S. Shen, W. Dai, L. Zhou, Z. Liu, and Y. Zhao, “Phantom: A cuda-accelerated word-wise homomorphic encryption library,” IEEE Trans. Dependable Secur. Comput., vol. 21, no. 5, pp. 4895–4906, 2024. [Online]. Available: https://doi.org/10.1109/ TDSC.2024.3363900

Listed on the reference page of the site.

SoK reference [93]

H. Yang, S. Shen, W. Dai, L. Zhou, Z. Liu, and Y. Zhao, “Phantom: A cuda-accelerated word-wise homomorphic encryption library,” IEEE Transactions on Dependable and Secure Computing, vol. 21, no. 5, pp. 4895–4906, 2024.

Listed on the reference page of the site.

Keyword mentions

Counts of vocabulary terms in the full text (first page of occurrence in brackets).

  • scheme: CKKS 45 [p.1], BGV 44 [p.1], BFV 40 [p.1], Leveled / somewhat HE 5 [p.1], FHEW 2 [p.1], RSA 2 [p.18]
  • library: Microsoft SEAL 8 [p.2], Phantom 6 [p.1], HElib 3 [p.2], OpenFHE 2 [p.2], PALISADE 2 [p.2]
  • model: Decision tree 1 [p.1]
  • dataset: MNIST 1 [p.2], ImageNet 1 [p.2]
  • hardware: GPU 48 [p.1], CUDA 5 [p.1], NVIDIA A100 4 [p.13], NVIDIA V100 1 [p.17], NVIDIA RTX 1 [p.16], FPGA 1 [p.2], Multi-GPU 1 [p.16]
  • technique: RNS 38 [p.1], Key switching 27 [p.1], Bootstrapping 7 [p.1], Rotations 7 [p.3], Rescaling 6 [p.3], Modulus switching 6 [p.6], NTT 6 [p.4], Relinearization 4 [p.3], Convolution 4 [p.3], SIMD packing 2 [p.2], Non-interactive 2 [p.1]
  • assumption: LWE 4 [p.1], Lattice-based 2 [p.18]
  • task: Language modeling / generation 4 [p.3]
  • topic: Homomorphic encryption 51 [p.1], Benchmark 7 [p.1], Open source 7 [p.1]

Files

Path (under papers/) Collection Pages Size SHA-256 Note
references/catalogue/7-097_phantom-a-cuda-accelerated-word-wise-homomorphic-encryption.pdf catalogue 18 939.4 KB ab8e85434f78
references/sok-references/088_phantom-a-cuda-accelerated-word-wise-homomorphic-encryption.pdf sok-references 18 939.4 KB ab8e85434f78 byte-identical to references/catalogue/7-097_phantom-a-cuda-accelerated-word-wise-homomorphic-encryption.pdf
references/sok-references/093_phantom-a-cuda-accelerated-word-wise-homomorphic-encryption.pdf sok-references 18 939.4 KB ab8e85434f78 byte-identical to references/catalogue/7-097_phantom-a-cuda-accelerated-word-wise-homomorphic-encryption.pdf

Database key: phantom-a-cuda-accelerated-word-wise-homomorphic-encryption-library · created 2026-11-05 · updated 2026-11-05 · edit with python3 tools/paperdb.py set phantom-a-cuda-accelerated-word-wise-homomorphic-encryption-library COLUMN VALUE