Hybrid elastic ARM&cloud HPC collaborative platform for generic tasks
Material type:
ArticleDescription: 1 archivo (1,5 MB)Subject(s): Online resources: Summary: Compute-heavy workloads are currently run on Hybrid HPC structures using x86 CPUs and GPUs from Intel, AMD, or NVidia, which have extremely high energy and financial costs. However, thanks to the incredible progress made on CPUs and GPUs based on the ARM architecture and their ubiquity in today’s mobile devices, it’s possible to conceive of a low-cost solution for our world’s data processing needs. Every year ARM-based mobile devices become more powerful, efficient, and come in ever smaller packages with ever growing storage. At the same time, smartphones waste these capabilities at night while they’re charging. This repre-sents billions of idle devices whose processing power is not being utilized. For that reason, the objective of this paper is to evaluate and develop a hybrid, distributed, scalable, and redundant platform that allows for the utilization of these idle devices through a cloud-based administration service. The system would allow for massive improvements in terms of efficiency and cost for com-pute-heavy workload. During the evaluation phase, we were able to establish sav-ings in power and cost significant enough to justify exploring it as a serious al-ternative to traditional computing architectures.
| Item type | Home library | Collection | Call number | URL | Status | Date due | Barcode | |
|---|---|---|---|---|---|---|---|---|
Capítulo de libro
|
Biblioteca de la Facultad de Informática | Biblioteca digital | A1187 (Browse shelf(Opens below)) | Link to resource | No corresponde |
Formato de archivo PDF. -- Este documento es producción intelectual de la Facultad de Informática - UNLP (Colección BIPA/Biblioteca)
Compute-heavy workloads are currently run on Hybrid HPC structures using x86 CPUs and GPUs from Intel, AMD, or NVidia, which have extremely high energy and financial costs. However, thanks to the incredible progress made on CPUs and GPUs based on the ARM architecture and their ubiquity in today’s mobile devices, it’s possible to conceive of a low-cost solution for our world’s data processing needs. Every year ARM-based mobile devices become more powerful, efficient, and come in ever smaller packages with ever growing storage. At the same time, smartphones waste these capabilities at night while they’re charging. This repre-sents billions of idle devices whose processing power is not being utilized. For that reason, the objective of this paper is to evaluate and develop a hybrid, distributed, scalable, and redundant platform that allows for the utilization of these idle devices through a cloud-based administration service. The system would allow for massive improvements in terms of efficiency and cost for com-pute-heavy workload. During the evaluation phase, we were able to establish sav-ings in power and cost significant enough to justify exploring it as a serious al-ternative to traditional computing architectures.
Conference Cloud Computing and Big Data (7ª : 2019 : La Plata, Argentina)