Big Data Analytics and Machine Learning Technologies for HPC Applications
High-performance computing (HPC) has long been pivotal for carrying out data and compute-intensive large-scale scientific simulations, analytic workloads for advanced scientific progress and product innovations, in turn making HPC infrastructure more esse
- PDF / 201,327 Bytes
- 14 Pages / 439.37 x 666.142 pts Page_size
- 47 Downloads / 241 Views
Abstract High-performance computing (HPC) has long been pivotal for carrying out data and compute-intensive large-scale scientific simulations, analytic workloads for advanced scientific progress and product innovations, in turn making HPC infrastructure more essential and valuable than ever. The convergence of HPC and big data technologies/machine learning (ML)/ deep learning (DL) is prescribed owing to the multifold growth of data across all HPC domains. HPC application experts are either already working or looking forward towards ML solutions to their applications, as it is proven successful in many scientific and commercial domains. Modern CPU/GPU-based HPC architectures are equipped with support for ML/DL promising the adaptability of AI capabilities in the HPC territory. Artificial intelligence (AI) enabled neuromorphic chips to add another ladder towards the convolution of AI on HPC. In this paper, we bring forth the merits of AI on HPC infrastructure for scientific applications in the shortlisted domains, viz weather and climate, astrophysics, agriculture, and bioinformatics. The paper discusses the current scenarios of wide adoption and merits of AI in the said domains. The survey lists the applications that are well received by the user communities for their performance, handling big unstructured data, improved results, etc., while solving the domain-specific problem. Keywords Artificial intelligence · Machine learning · Deep learning · High-performance computing · Weather and climate · Astrophysics · Agriculture · Bioinformatics
Sukeshini · P. Sharma (B) · M. Ved · J. Chintalapti · S. N. Pal Centre for Development of Advanced Computing, Bangalore, India e-mail: [email protected] M. Ved e-mail: [email protected] J. Chintalapti e-mail: [email protected] S. N. Pal e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 P. K. Singh et al. (eds.), Evolving Technologies for Computing, Communication and Smart World, Lecture Notes in Electrical Engineering 694, https://doi.org/10.1007/978-981-15-7804-5_31
411
412
Sukeshini et al.
1 Introduction HPC [1] is amalgamation of computing power delivering higher speed and performance for solving grand challenges and large-scale simulations in various scientific domains and for faster businesses solutions. The well-established de-facto technology, HPC has been embraced by government agencies, academics, enterprises, and scientific researchers’ communities for economic viability to innovate breakthrough product and services. Be it the vertically scaled up supercomputers or the horizontally scaled-out, clusters of servers running in parallel to create supercomputing-class throughput, HPC has spread across enterprises. HPC researchers and developers can focus on being productive by quicker iterations of their empirical work, when they break free from the cycle of concept → design → code → execute → result. The usage and impact of HPC touches almost every facet of daily life. Big data has moved way past the initial V’s iden
Data Loading...
