English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Fully Homomorphic Encryption (FHE) allows arbitrarily complex computations onencrypted data without ever needing to decrypt it, thus enabling us to maintaindata privacy on third-party systems. Unfortunately, sustaining deepcomputations with FHE requires a periodic noise reduction step known asbootstrapping. The cost of the bootstrapping operation is one of the primarybarriers to the wide-spread adoption of FHE. In this paper, we present anin-depth architectural analysis of the bootstrapping step in FHE. First, weobserve that secure implementations of bootstrapping exhibit a low arithmeticintensity (<1 Op/byte), require large caches (>100 MB), and are heavily boundby the main memory bandwidth. Consequently, we demonstrate that existingworkloads observe marginal performance gains from the design of bespokehigh-throughput arithmetic units tailored to FHE. Second, we propose severalcache-friendly algorithmic optimizations that improve the throughput in FHEbootstrapping by enabling up to 3.2x higher arithmetic intensity and 4.6x lowermemory bandwidth. Our optimizations apply to a wide range of structurallysimilar computations such as private evaluation and training of machinelearning models. Finally, we incorporate these optimizations into anarchitectural tool which, given a cache size, memory subsystem, the number offunctional units and a desired security level, selects optimal cryptosystemparameters to maximize the bootstrapping throughput. Our optimizedbootstrapping implementation represents a best-case scenario for computeacceleration of FHE. We show that despite these optimizations, bootstrappingcontinues to be bottlenecked by main memory bandwidth. We propose new researchdirections to address the underlying memory bottleneck. In summary, our answerto the titular question is: yes, but only after addressing the memorybottleneck!

Does Fully Homomorphic Encryption Need Compute Acceleration?