Paper Info
Title
Aquabolt-XL: Samsung HBM2-PIM with in-memory processing for ML accelerators and beyond
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.
Year
DOI
Venue
2021
10.1109/HCS52781.2021.9567191
2021 IEEE Hot Chips 33 Symposium (HCS)
Keywords
DocType
ISBN
Wires,Memory management,Bandwidth,Energy efficiency
Conference
978-1-6654-1397-8
Citations 
PageRank 
References 
2
0.40
0
Authors
20
Name
Order
Citations
PageRank
Jin Hyun Kim19221.61
Shinhaeng Kang2192.04
Sukhan Lee331.43
Hyeonsu Kim420.40
Woongjae Song520.40
Yuhwan Ro620.40
Seungwon Lee712732.51
David Wang8181.69
Hyunsung Shin9442.86
Bengseng Phuah1020.40
Jihyun Choi1120.40
Jinin So1220.40
YeonGon Cho1320.40
JoonHo Song1420.40
Jangseok Choi1520.40
Jeong-Hyeon Cho1631.77
Kyomin Sohn1720.40
Young-Soo Sohn1811521.21
Kwang-Il Park1916325.68
Nam Sung Kim203268225.99