Abstract | ||
---|---|---|
Memory has become the bottleneck of whole system as the processing data size greatly increases, motivating memory industry moving forward. Hybrid Memory Cube (HMC) can reach much shorter transaction latency and higher bandwidth, as well as providing a fascinating feature named chaining. Chaining technology can integrate multiple HMCs together to obtain larger capacity while keeping the benefit of latency and bandwidth. In this paper we proposed a mesh HMC network and made a comprehensive study on it. We learned that compared with single HMC, the mesh HMC network can have much larger capacity and lower power consumption for each HMC while only incurring slight latency increment. |
Year | DOI | Venue |
---|---|---|
2020 | 10.1109/ISPASS48437.2020.00038 | 2020 IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS) |
Keywords | DocType | ISBN |
Hybrid Memory Cube,Chaining,Network | Conference | 978-1-7281-4799-4 |
Citations | PageRank | References |
0 | 0.34 | 3 |
Authors | ||
4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Cheng Qian | 1 | 13 | 8.11 |
Ming Zhang | 2 | 89 | 18.62 |
Xiaohui Kuang | 3 | 17 | 9.61 |
Gang Zhao | 4 | 20 | 8.59 |