Abstract
As Moore's Law nears its end, we are searching for alternative technologies and architectures to further increase performance. Cryogenic computing has gained considerable attention in the last couple years, due to the highly ideal performance of CMOS circuits at very low temperatures. While cryogenic operation can provide impressive performance benefits, it introduces a new trade-off space which must be examined. Additionally, it does not eliminate current bottlenecks for performance, such as the memory wall. Processing-in-Memory architectures present an interesting opportunity. They are suitable to operate at cryogenic temperatures, enable lower cooling costs via extreme energy efficiency, and enable compute and memory capabilities in this regime with relatively minor adjustments to the architecture.
Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
