Reproduce “RowHammer Vulnerability Counter (RVC): Redefining RowHammer Detection with Victim-Centric Tracking” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Lavi Jain, Venkata Kalyan Tavva
The RowHammer Vulnerability Counter (RVC) framework shifts from aggressor-centric to victim-centric detection, monitoring a row's direct vulnerability to bit flips. This approach achieves an average 92.3% reduction in refreshes, 91.3% less VRR-induced energy, and a 5.5% improvement in LLC latency compared to Graphene, without increasing hardware overhead.
Key insights
- A victim-centric approach to RowHammer detection, focusing on actual row vulnerability, drastically reduces redundant refresh operations.
- Accessing a memory row inherently refreshes it, making explicit refresh actions for already accessed rows unnecessary.
- The cumulative impact of multiple aggressors on a victim row can be effectively accounted for by directly incrementing the victim's counter, simplifying threshold calculations.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.