Coursera Spring Sale
40% Off Coursera Plus Annual!
Grab it
Explore a groundbreaking 15-minute conference presentation that introduces the first thin-air free memory model capable of supporting aggressive compiler optimizations while maintaining semantic correctness. Learn how this innovative approach leverages alias analysis knowledge, assumes freedom from undefined behavior, and accommodates extrinsic implementation choices like over-alignment in real-world systems. Discover the model's state-of-the-art tooling support that executes test batteries orders of magnitude faster than other executable thin-air free semantics, making it highly practical for development workflows. Understand how the model seamlessly integrates with the C/C++ memory model through an exportable semantic dependency relation, supports standard compilation mappings for atomics, and successfully passes all tests from the recently published ISO desiderata for C/C++. Gain insights into how this research addresses the long-standing thin-air problem in relaxed memory models while enabling the compiler optimizations that modern high-performance systems require, presented by researchers from the University of Kent, University of Surrey, and NVIDIA at the OOPSLA 2025 conference.
