14:00 – 14:20

Welcome Talk

Ingrid Verbauwhede – Belfort – COSIC

14:20 – 15:00

Waiting for BASALISC: Silicon in the oven!

David Archer – Niobium

Fully Homomorphic Encryption (FHE) provably assures confidentiality of sensitive data during computation. Unfortunately, high computational cost and excessive memory bandwidth requirements prevent its adoption today in most settings. We present BASALISC, the first in a family of hardware coprocessors that mitigate those drawbacks and accelerate FHE computations. BASALISC has been described elsewhere in preliminary form. This paper provides an overview of BASALISC now that its design, validation, and tape-out are complete and manufacturing is underway. BASALISC is the first FHE hardware solution to implement both the BGV and CKKS FHE schemes with fully-packed bootstrapping in ASIC form. BASALISC is a RISC architecture with a small instruction set and a single data type. BASALISC taped out as a 1 GHz ASIC on a 12nm Global Foundries semiconductor process with a die size of 13.4mm x 20mm. Simulation results show a speedup over software-based homomorphic encryption of several thousand times for linear regression and neural network inference applications.

15:00 – 15:40

The Hitchhiker Guide to EC Adder Hardware Acceleration

Yuval Domb – Ingonyama

15:40 – 16:00

Coffee Break

16:00 – 16:40

Making FHE hardware-acceleration-friendly (and vice versa)

Michiel Van Beirendonck – Belfort / COSIC

We will present our newest Belfort accelerator, ready to accelerate TFHE applications for any parameter set. We will shed a light on the current state of acceleration, and on the advances to come.

16:40 – 17:00

Elevator Pitch Session

17:00 – 20:00

Reception + Walking dinner with demo’s & posters

8:30 – 9:00

Welcome Coffee

9:00 – 9:40

Hardware Acceleration of FHE using the power of light: Introducing the Optalysys Explore Platform

Nick New – Optalysys

Optalysys are developing their unique photonic approach to accelerate the base functions which dominate the processing of data encrypted under FHE. In this talk Dr New will present the roadmap of their Explore platform, which allows users to access the hardware via the main libraries on the cloud.

9:40 – 10:20

Intel HERACLES

Rosario Cammarota – Intel

Intel HERACLES introduces a groundbreaking near-memory computer architecture optimized for Fully Homomorphic Encryption (FHE) programs, leveraging native processing of ring polynomials. Its dedicated software stack, designed to harness FHE programs’ unique characteristics, efficiently schedules tasks on HERACLES, utilizing advanced FHE algorithms and hardware-assisted features to minimize metadata and evaluation key data movement. This approach significantly enhances performance, achieving up to four orders of magnitude speedup compared to contemporary data center CPUs. HERACLES’ compact form factor, akin to standard data center accelerators, enables a paradigm shift in FHE application performance, moving from batch to close to real-time processing, opening new avenues for efficient encrypted data processing across various industries.

10:20 – 11:00

Coffee Break

11:00 – 11:20

Confidential smart contracts using homomorphic encryption

Jason Delabays – Zama

In this presentation, we will explore applications of Fully Homomorphic Encryption (FHE) within blockchain technology, focusing on asset tokenization which requires enhanced confidentiality. We will provide a clear overview of the architecture of fhEVM, our Ethereum Virtual Machine enhanced with FHE capabilities, which allows for encrypted state computations and secure, private transactions. Additionally, we will cover the fundamental principles of FHE, highlighting its unique suitability for complex operations in financial and blockchain environments. This session aims to shed light on how fhEVM can revolutionize blockchain use cases by ensuring privacy without compromising on functionality.

11:20 – 11:40

Cornamic Confidential Compute

Paul Master – Cornami

11:40 – 12:20

An NVMe-based Secure Computing Platform with FPGA-based TFHE Accelerator

Yoshihiro Ohba – Kioxia

We design and implement an NVMe-based secure computing platform with an FPGA-based Torus FHE (TFHE) accelerator, SSD, and a host-sided middleware. We provide a full-fledged secure computing platform for TFHE using an FPGA-based accelerator. We define a set of secure computing instructions to evaluate any 14-bit to 14-bit function using TFHE. Our secure computing platform outperforms a software-based platform with a single high-end CPU by 15 times in gate bootstrapping execution time and by 12 times in electric energy consumption during the gate bootstrapping execution time.

12:30 – 14:00

Lunch

14:10 – 14:50

Building & Standardizing an Ecosystem for Encrypted Computation with MLIR

Alexander Viand – Intel

14:50 – 15:30

HEIR: A unified compiler for FHE with extensible backends

Shruthi Gorantala & Asra Ali – Google

15:30 – 16:00

Coffee

16:00 – 16:40

Programming Performant ZK Proofs — Accelerating ZK through AI Compute Paradigms

Thomas De Cnudde – Fabric Cryptography

16:40 – 17:20

Final Pitch + Discussions + Future work