Formal Verification of a Custom Compiler for a Fully Homomorphic Encryption Accelerator

We present a methodology to formally verify a custom compiler for a Fully Homomorphic Encryption (FHE) accelerator. Prior research has addressed formal verification of the hardware component of such accelerators. Equally important is the correctness of the software stack that decomposes and com-piles an FHE workload into the actual sequence of instructions that execute on the hardware. To ensure the correctness of the resulting instruction sequences, we apply a three-step verification process to 1) check for deadlocks, 2) build a baseline functional model of the instructions, and 3) verify that all possible executions of the instructions, subject to hardware design constraints such as latency variability, concurrency, etc. on the accelerator, will always match the baseline model. Coupling this verification step with the compiler not only guarantees correctness of the compiler results, but also enables quick feedback on the correctness of experimental optimizations as the compiler continues to evolve. We demonstrate how the proposed methodology is capable of identifying different types of bugs in a FHE accelerator compiler implementation.

Authors

Institutions

Publication Details

Published
2025-04-23
DOI
https://doi.org/10.1109/isqed65160.2025.11014422
Primary Topic
Cryptographic Implementations and Security
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

Formal Verification of a Custom Compiler for a Fully Homomorphic Encryption Accelerator

Zhenkun Yang, Jin Yang, Jeremy Casas, Suvadeep Banerjee
Cryptographic Implementations and Security
article

Formal Verification of a Custom Compiler for a Fully Homomorphic Encryption Accelerator

Zhenkun Yang, Jin Yang, Jeremy Casas, Suvadeep Banerjee
article en

Abstract

We present a methodology to formally verify a custom compiler for a Fully Homomorphic Encryption (FHE) accelerator. Prior research has addressed formal verification of the hardware component of such accelerators. Equally important is the correctness of the software stack that decomposes and com-piles an FHE workload into the actual sequence of instructions that execute on the hardware. To ensure the correctness of the resulting instruction sequences, we apply a three-step verification process to 1) check for deadlocks, 2) build a baseline functional model of the instructions, and 3) verify that all possible executions of the instructions, subject to hardware design constraints such as latency variability, concurrency, etc. on the accelerator, will always match the baseline model. Coupling this verification step with the compiler not only guarantees correctness of the compiler results, but also enables quick feedback on the correctness of experimental optimizations as the compiler continues to evolve. We demonstrate how the proposed methodology is capable of identifying different types of bugs in a FHE accelerator compiler implementation.

Intel (United States) (US)
Openalex Percentile: Top 94%
Cryptographic Implementations and Security
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.