Skip to content

BASALISC: Programmable Hardware Accelerator for BGV Fully Homomorphic Encryption

Authors: Robin Geelen; Michiel Van Beirendonck; Hilder V. L. Pereira; Brian Huffman; Tynan McAuley; Ben Selfridge; Daniel Wagner; Georgios Dimou; Ingrid Verbauwhede; Frederik Vercauteren; David W. Archer
Year: 2023
Venue: USENIX Security 2023

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

Links: arXiv:2205.14017 · ePrint 2022/657

Summary

Auto-extracted summary (unreviewed)

BASALISC is a family of programmable RISC-style ASIC accelerators for BGV fully homomorphic encryption and the first to support fully-packed BGV bootstrapping in hardware. Designed for 1 GHz in a 12 nm GlobalFoundries process on a 150 mm2 die, it combines a four-layer memory hierarchy with a conflict-free inner memory enabling 32 Tb/s radix-256 NTTs, permutation hardware reused for automorphisms, a custom key-switching MAC unit, and a compiler plus performance/correctness simulator. Timing closure, formal verification of the functional units and simulation on three macro-benchmarks show more than 5,000x speedup over HElib.

Abstract

Fully Homomorphic Encryption (FHE) allows for secure computation on encrypted data. Unfortunately, huge memory size, computational cost and bandwidth requirements limit its practicality. We present BASALISC, an architecture family of hardware accelerators that aims to substantially accelerate FHE computations in the cloud. BASALISC is the first to implement the BGV scheme with fully-packed bootstrapping -- the noise removal capability necessary for arbitrary-depth computation. It supports a customized version of bootstrapping that can be instantiated with hardware multipliers optimized for area and power. BASALISC is a three-abstraction-layer RISC architecture, designed for a 1 GHz ASIC implementation and underway toward 150mm2 die tape-out in a 12nm GF process. BASALISC's four-layer memory hierarchy includes a two-dimensional conflict-free inner memory layer that enables 32 Tb/s radix-256 NTT computations without pipeline stalls. Its conflict-resolution permutation hardware is generalized and re-used to compute BGV automorphisms without throughput penalty. BASALISC also has a custom multiply-accumulate unit to accelerate BGV key switching. The BASALISC toolchain comprises a custom compiler and a joint performance and correctness simulator. To evaluate BASALISC, we study its physical realizability, emulate and formally verify its core functional units, and we study its performance on a set of benchmarks. Simulation results show a speedup of more than 5,000 times over HElib -- a popular software FHE library.
Source: arxiv

Extracted details

Field Value Source Confidence
contribution first hardware architecture implementing BGV with fully-packed bootstrapping llm 0.85
contribution three-abstraction-layer RISC ISA and four-layer memory hierarchy with conflict-free 2D inner memory llm 0.85
contribution radix-256 NTT at 32 Tb/s without pipeline stalls llm 0.85
contribution permutation hardware generalized to compute BGV automorphisms at no throughput penalty llm 0.85
contribution custom multiply-accumulate unit for key switching and multipliers restricted to NTT-friendly primes llm 0.85
contribution custom compiler and joint performance/correctness simulator llm 0.85
contribution physical realizability study, emulation and formal verification of core units llm 0.85
schemes BGV llm 0.85
libraries HElib llm 0.85
tasks FHE acceleration llm 0.85
tasks bootstrapping llm 0.85
hardware ASIC, 12 nm GlobalFoundries, 1 GHz, 150 mm2 (simulated, not yet fabricated) llm 0.85
security level 128-bit llm 0.85
interaction non-interactive FHE llm 0.85
code url https://github.com/homenc/HElib/tree/master/examples/BGV_country_db_lookup (URL found in the full text) regex 0.50
code url https://github.com/tuneinsight/lattigo (URL found in the full text) regex 0.50
code url https://github.com/Microsoft/SEAL (URL found in the full text) regex 0.50
limitations closed-source architecture; implementation in progress and not yet delivered to silicon, so evaluation relies on simulation and emulation llm 0.85
limitations multipliers restricted to NTT-friendly primes llm 0.85
llm relevance low llm 0.85
key result speedup over HElib: more than 5,000x (simulation on three FHE macro-benchmarks) llm 0.85
key result NTT throughput: 32 Tb/s radix-256 (conflict-free inner memory layer) llm 0.85
key result timing closure: 1.0 GHz at SS corner, 0.72 V (12 nm GF low-power process, PEs and CTB RAM placed and routed) llm 0.85
key result die size: 14.4 mm x 10.4 mm (target 150 mm2) (floorplan with DDR4 and PCIe macros) llm 0.85

System card: BASALISC

Group 8. FHE hardware accelerators
Family ASIC (BGV)
Arithmetic -
Interaction Hardware
Model scale -
Status Active
Relation to the SoK BGV accelerator (DARPA DPRIVE).
Note Exact-arithmetic counterpart to CKKS accelerators.
Systems page BASALISC

SoK reference [105]

R. Geelen, M. Van Beirendonck, H. V. L. Pereira, B. Huffman, T. McAuley, B. Selfridge, D. Wagner, G. D. Dimou, I. Verbauwhede, F. Vercauteren, and D. W. Archer, “BASALISC: Programmable hardware accelerator for BGV fully homomorphic encryption,” IACR Transactions on Cryptographic Hardware and Embedded Systems, vol. 2023, no. 4, pp. 32–57, 2023.

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: BGV 52 [p.1], CKKS 12 [p.20], MPC 3 [p.1], Leveled / somewhat HE 3 [p.22]
  • library: HElib 19 [p.1], Microsoft SEAL 3 [p.20], Lattigo 2 [p.20], Concrete (Zama) 2 [p.23]
  • model: Logistic regression 9 [p.3]
  • hardware: ASIC 6 [p.1], FPGA 3 [p.26], Intel Xeon 1 [p.18], GPU 1 [p.2]
  • technique: NTT 93 [p.1], Bootstrapping 63 [p.1], Key switching 32 [p.1], Compiler 9 [p.1], RNS 8 [p.5], Modulus switching 5 [p.3], SIMD packing 2 [p.6], Convolution 2 [p.6]
  • assumption: RLWE 5 [p.22], 128-bit security 3 [p.6], LWE 2 [p.23], Lattice-based 2 [p.23]
  • topic: Homomorphic encryption 68 [p.1], Benchmark 16 [p.1], Open source 3 [p.24]

Files

Path (under papers/) Collection Pages Size SHA-256 Note
references/catalogue/8-118_basalisc-programmable-hardware-accelerator-for-bgv-fully-hom.pdf catalogue 26 3.2 MB 088b1f7fa3df
references/sok-references/105_basalisc-programmable-hardware-accelerator-for-bgv-fully-hom.pdf sok-references 26 3.4 MB 35129b622cd7

Database key: basalisc-programmable-hardware-accelerator-for-bgv-fully-homomorphic · created 2026-11-05 · updated 2026-11-05 · edit with python3 tools/paperdb.py set basalisc-programmable-hardware-accelerator-for-bgv-fully-homomorphic COLUMN VALUE