GPUs and CPUs Issue Warnings
NVIDIA warns of GPUHammer attacks targeting AI models while AMD discovers new CPU vulnerabilities. Learn about these critical hardware security threats and protection methods.
"ChipPub" Publication: 20% Discount Offer Link.
NVIDIA is urging customers to enable system-level Error Correcting Code (ECC) to defend against RowHammer attack variants targeting its Graphics Processing Units (GPUs).
The GPU manufacturer stated in an advisory released this week: "The risk of successful RowHammer attack exploitation varies depending on DRAM devices, platforms, design specifications, and system configurations."
This attack, dubbed GPUHammer, represents the first RowHammer vulnerability exploitation targeting NVIDIA GPUs (such as NVIDIA A6000 GPUs with GDDR6 memory), enabling malicious GPU users to tamper with other users' data by triggering bit flips in GPU memory.
Researchers from the University of Toronto discovered that the most concerning consequence of this behavior is the degradation of artificial intelligence (AI) model accuracy from 80% to below 1%.
RowHammer is to modern DRAM what Spectre and Meltdown are to contemporary CPUs. While both are hardware-level security vulnerabilities, RowHammer targets the physical behavior of DRAM memory, whereas Spectre exploits speculative execution in CPUs.
RowHammer causes bit flips in nearby memory cells due to electrical interference in DRAM resulting from repeated memory access, while Spectre and Meltdown allow attackers to obtain privileged information from memory through side-channel attacks, potentially leaking sensitive data.