Overview
Save Big on Coursera Plus. 7,000+ courses at $160 off. Limited Time Only!
Explore the security implications of integrated Graphical Processor Units (GPUs) in commodity processors through this IEEE conference talk. Delve into how GPUs, typically used for benign applications like image rendering, can be exploited to accelerate microarchitectural attacks on common platforms. Learn about the construction of GPU-based primitives for effective microarchitectural attacks, their exposure through standardized browser extensions, and their ability to bypass state-of-the-art mitigations. Discover novel reverse engineering techniques used to uncover the cache architecture and replacement policy of the Adreno 330 GPU. Gain insights into the first end-to-end microarchitectural compromise of a mobile phone browser in under two minutes using orchestrated GPU primitives. Examine the challenges in implementing these attacks due to undocumented hardware features and discuss potential mitigations against GPU-enabled attackers.
Syllabus
Grand Pwning Unit: Accelerating Microarchitectural Attacks with the GPU
Taught by
IEEE Symposium on Security and Privacy