Tier-2 Optimisation for Computational Density/Diversity and Big Data

20th International Conference on Computing in High Energy and Nuclear Physics

物理学;计算机科学

Fay, R.B.^1 ; Bland, J.^1

Department of Physics, University of Liverpool, Liverpool

L69 7ZE, United Kingdom^1

关键词: Computational density;    Computational power;    Emerging technologies;    Graphical processing units;    Heterogeneous platforms;    Multi-core systems;    RAID controllers;    Solid state drives (SSD);   

Others  :  https://iopscience.iop.org/article/10.1088/1742-6596/513/6/062012/pdf
DOI  :  10.1088/1742-6596/513/6/062012

学科分类：计算机科学（综合）

来源: IOP

【 摘 要 】

As the number of cores on chip continues to trend upwards and new CPU architectures emerge, increasing CPU density and diversity presents multiple challenges to site administrators. These include scheduling for massively multi-core systems (potentially including Graphical Processing Units (GPU), integrated and dedicated) and Many Integrated Core (MIC)) to ensure a balanced throughput of jobs while preserving overall cluster throughput, as well as the increasing complexity of developing for these heterogeneous platforms, and the challenge in managing this more complex mix of resources. In addition, meeting data demands as both dataset sizes increase and as the rate of demand scales with increased computational power requires additional performance from the associated storage elements. In this report, we evaluate one emerging technology, Solid State Drive (SSD) caching for RAID controllers, with consideration to its potential to assist in meeting evolving demand. We also briefly consider the broader developing trends outlined above in order to identify issues that may develop and assess what actions should be taken in the immediate term to address those.