EVAX paper proposes on-demand defenses for microarchitectural attacks
Original title: EVAX: Towards a Practical, Pro-active & Adaptive Architecture for High Performance & Security
Authors report cutting InvisiSpec overhead from 27% to 1.26% and Fencing from 74% to 3.45%.
An arXiv paper describes EVAX, a generative training framework and detector meant to activate microarchitectural defenses only when attacks are detected. The authors say it covers Spectre-class, LVI, and Rowhammer attacks plus some unknown variants, with four false positives per million instructions in their experiments. The work is a research proposal, not a shipped processor feature.
Unconfirmed: Overhead and detection figures are the authors' experimental claims and are not independently confirmed here.
- Stated overhead falls to 1.26% for InvisiSpec and 3.45% for Fencing.
- The detector reported four false positives per million instructions in the authors' tests.
- Defenses stay off until the detector flags an attack.
From the original article
1 min read at arxiv.orgThis paper provides an end-to-end solution to defend against known microarchitectural attacks such as speculative execution attacks, fault-injection attacks, covert and side channel attacks, and unknown or evasive versions of these attacks. Current defenses are attack specific and can have unacceptably high performance overhead.
By Samira Mirbagher Ajorpaz · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.