Accelerating fully homomorphic encryption using GPU

By Wei Wang, Yin Hu, Lianmu Chen, Xinming Huang, Berk Sunar. 2012

135 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Information Systems

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Information Systems

Citations in the corpus

HEAX: An Architecture for Computing on Encrypted Data (2019, methodology): “Similarly, we leveraged the RNS-level parallelism which is proven to be secure [16].”
High-Speed Fully Homomorphic Encryption Over the Integers (2014, methodology): “Large-integer FFT multiplication has also been used in the previously mentioned hardware and GPU implementations of other FHE schemes.”
Accelerating NTRU based homomorphic encryption using GPUs (2014, methodology): “To tackle the efficiency bottleneck the use of alternative platforms, such as GPUs [16], reconfigurable logic [17]–[21] and custom ASIC [22], [23] has been proposed.”
cuHE: A Homomorphic Encryption Accelerator Library (2015, background): “The results of [35, 9, 10] demonstrate the power of GPU-accelerated HE evaluations.”
SHIELD: Scalable Homomorphic Implementation of Encrypted Data-Classifiers (2015, background): “In the testing phase however, the secure email spam filter is required to classify the input encrypted email as a spam or not without any knowledge about the actual contents of the email.”
Accelerating LTV Based Homomorphic Encryption in Reconfigurable Hardware (2015, methodology): “Therefore, a good number of works focused on developing FFT/NTT based large integer multipliers [17,18,35,35].”
Modular Hardware Architecture for Somewhat Homomorphic Function Evaluation (2015, methodology): “Wang et al. [38] implemented primitives for the Gentry-Halevi (GH) FHE scheme [23] using Graphics Processing Unit (GPU) and observed speedups up to 7.68.”
FPGA implementation of a large-number multiplier for fully homomorphic encryption (2013, methodology): “4 times faster than the original implemenation on an Intel Xeon processor after we did further optimization on [3].”
Accelerating integer-based fully homomorphic encryption using Comba multiplication (2014, background): “Thus, to address this shortcoming, optimised architectures that target alternative platforms, such as Graphics Processing Units (GPUs) and FPGAs, have recently been proposed [14]– [19].”
Accelerating Fully Homomorphic Encryption in Hardware (2015, methodology): “1 GPU [14] 0.”
Accelerating Fully Homomorphic Encryption over the Integers with Super-size Hardware Multiplier and Modular Reduction (2013, methodology): “The most recent software implementation was done on a NVIDIA C2050 GPU [13], and it uses the integer-FFT multiplication algorithm [14] to compute the super-size multiplication and Barrett modular reduction [15] to implement Gentry’s FHE scheme [2].”
Accelerating leveled fully homomorphic encryption using GPU (2014, methodology): “We took an initial step to accelerate Gentry-Halevi scheme using the integer-FFT multiplication algorithm on NVIDIA C2050 GPU [7], which resulted about 7 times speedup compared the original work in [6].”
A Custom Accelerator for Homomorphic Encryption Applications (2017, methodology): “[28] managed to accelerate the recryption primitive of Gentry and Halevi [26] by roughly an order of magnitude.”
HEAX: High-Performance Architecture for Computation on Homomorphically Encrypted Data in the Cloud (2019, methodology): “[72] have proposed the first GPU acceleration of FHE that targets Gentry-Halevi [35] scheme.”
Accelerating Homomorphic Evaluation on Reconfigurable Hardware (2015, methodology): “[42] present the first GPU implementation of an FHE scheme and provide results for the Gentry-Halevi [24] scheme on an NVIDIA C2050 GPU.”
Accelerating Somewhat Homomorphic Evaluation using FPGAs (2015, methodology): “765 ms NTT mul [35] GH-FHE NVIDIA GTX 690 0.”
F1: A Fast and Programmable Accelerator for Fully Homomorphic Encryption (2021, methodology): “GPU acceleration: Finally, prior work has also used GPUs to accelerate different FHE schemes, including GH [74, 75], BGV [73], and B/FV [1].”