计算机科学
功率消耗
带宽(计算)
解算器
功率(物理)
电信
物理
程序设计语言
量子力学
作者
Keeyoung Son,Joonsang Park,Seongguk Kim,Boogyo Sim,Keunwoo Kim,Seonguk Choi,Hyun-Sik Kim,Joungho Kim
标识
DOI:10.1109/edaps58880.2023.10468315
摘要
In this paper, thermal analysis of high bandwidth memory (HBM)-GPU module considering power consumption was carried out. As the demand of high performance computing is increases, the importance of HBM is arisen due to serve high bandwidth and capacity with low power consumption. However, as HBM generation is developed, the power consumption of HBM are increases for high speed and performance. Therefore, these technical trends demand the thermal analysis of HBM module for thermally reliable system. Hence, this research analyzed the HBM-GPU module with various HBM power consumption. By using 3D fluent solver, we analyzed the junction temperature of HBM depending on power consumption of HBM, and effects by each heat generation sources in HBM-GPU module. The results verified that the main thermal issue is the GPU thermal coupling when the HBM power consumption is less than 30 W, and the HBM PHY self-generating heat when it exceeds.
科研通智能强力驱动
Strongly Powered by AbleSci AI