Aquabolt-XL: Samsung HBM2-PIM with in-memory processing for ML accelerators and beyond
Jin Hyun Kim, Shin-haeng Kang, Sukhan Lee, Hyeonsu Kim, Woongjae Song, Yuhwan Ro, Seungwon Lee, David Wang, Hyunsung Shin, Bengseng Phuah, Ji-Hyun Choi, Jinin So, Yeongon Cho, Joonho Song, Jangseok Choi, Jeong‐Hyeon Cho, Kyomin Sohn, Young‐Soo Sohn, Kwang‐Il Park, Nam Sung Kim
Abstract
Using PIM to overcome memory bottleneck • Although various bandwidth increase methods have been proposed, it is physically impossible to achieve a breakthrough increase. - Limited by # of PCB wires, # of CPU ball, and thermal constraints • PIM has been proposed to improve performance of bandwidth-intensive workloads and improve energy efficiency by reducing computing-memory data movement.
Topics & Concepts
BottleneckComputer scienceBandwidth (computing)Memory bandwidthEfficient energy useEmbedded systemParallel computingComputer architectureOperating systemElectrical engineeringEngineeringComputer networkAdvancements in Semiconductor Devices and Circuit DesignAdvanced Data Storage TechnologiesSuperconducting Materials and Applications