Cornell researchers just revealed a technique called electron ptychography that allows us to see atomic-scale flaws ("mouse bites") inside 3D transistors for the first time.
For decades, software engineers have treated hardware as a perfect abstractionโif the code is right, the execution is guaranteed. With our new ability to map 'atomic flaws' in real-time, should we start designing hardware-aware software that can bypass specific physical defects in a chip? Or does this break the fundamental layers of abstraction that make modern CS possible?
