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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

GPU clouds have become a popular computing platform because of the cost ofowning and maintaining high-performance computing clusters. Many cloudarchitectures have also been proposed to ensure a secure execution environmentfor guest applications by enforcing strong security policies to isolate theuntrusted hypervisor from the guest virtual machines (VMs). In this paper, westudy the impact of GPU chip's hardware faults on the security of cloud"trusted" execution environment using Deep Neural Network (DNN) as theunderlying application. We show that transient hardware faults of GPUs can begenerated by exploiting the Dynamic Voltage and Frequency Scaling (DVFS)technology, and these faults may cause computation errors, but they havelimited impact on the inference accuracy of DNN due to the robustness andfault-tolerant nature of well-developed DNN models. To take full advantage ofthese transient hardware faults, we propose the Lightning attack to locate thefault injection targets of DNNs and to control the fault injection precision interms of timing and position. We conduct experiments on three commodity GPUs toattack four widely-used DNNs. Experimental results show that the proposedattack can reduce the inference accuracy of the models by as high as 78.3\% and64.5\% on average. More importantly, 67.9\% of the targeted attacks havesuccessfully misled the models to give our desired incorrect inference result.This demonstrates that the secure isolation on GPU clouds is vulnerable againsttransient hardware faults and the computation results may not be trusted.

Lightning: Striking the Secure Isolation on GPU Clouds with Transient Hardware Faults