Table 6 Comparison of acceleration schemes based on GPU
References | Year | Availability | Data storage optimization | Homomorphic parameters tuning | Accelerated object | Acceleration result (times) |
|---|---|---|---|---|---|---|
Ahmad Al Badawi (2019) | 2019 | + | × | × | HPS RNS variant of the BFV | * |
Lupascu (2019) | 2019 | + + + | × | × | Ciphertext multiplication | 5.14 × |
Lei et al. (2019) | 2019 | + + + | × | × | KeyGen / Bootstrapping | 1.672–13.268 × |
Xia et al. (2019) | 2019 | + + | × | × | DGHV | 1.67–1.84 × |
Kim and Jung (2020) | 2020 | + + + | √ | × | NTT | 6.52 × |
Morshed (2020) | 2020 | + + | × | × | TFHE | 14.5–20 × |
Ahmad Al Badawi (2020) | 2020 | + | × | × | CKKS | 10–100 × |
Ahmad Al Badawi (2021) | 2020 | + + | × | × | HPS RNS variant of the BFV | 10–1000 × |
Pedro et al. (2021) | 2021 | + + + | × | × | BFV | 2.6 × |
Goey et al. 2021) | 2021 | + + | √ | × | CMNT | 1.41 × |
Jung et al. 2021) | 2021 | + + | √ | × | CKKS | 4.05 × |
Jung and Kim (2021) | 2021 | + + + + | √ | × | Bootstrapping | 7.02 × |
Castro et al. (2021) | 2021 | + | × | × | NTT / Polynomial Multiplication | 1092.29 × |
Özerk and Elgezen (2022) | 2022 | + + + | √ | × | NTT / Polynomial Multiplication | 90.13–141.95 × |
Türkoglu et al. (2022) | 2022 | + + + | × | × | HMult / Relinearization / Rotation / HAdd | 13.39–47.01 × |
Shivdikar et al. (2022) | 2022 | + + + + | √ | × | Polynomial Multiplication | 123.13 × |
Shen et al. (2022) | 2022 | + + + + | × | × | BGV / BFV / CKKS | 234.5–378.4 × |