Research
Projects
High Performance and Cloud Computing for high interest applications

High Performance and Cloud Computing for high interest applications

This project proposes different hardware and software solutions associated with high performance computing that provides a solution for the efficient and safe execution of applications that are challenging because of their complexity and that are demanded by priority sectors in the research and transfer strategy at European level. The development of solutions associated with manycore and multicore architectures as well as systems based on accelerators such as GPUs or FPGAs is proposed. Also, this project proposes solutions focused on Cloud systems, whose use is intensifying thanks to the popularity of the Internet of Objects (IO) and models such as Fog Computing. The tasks are organized into two main objectives: mass computing for challenging applications, and HPC and Cloud for processing large volumes of data.

This proposal is highly interdisciplinary as it combines aspects of research in the field of computer architecture and HPC with applications of special interest and in which the team has proven experience.

Objectives

Regarding the first objective, techniques and software solutions will be developed to different problems. On the one hand, techniques andtools will be proposed in the field of modeling and performance improvement both in terms of execution time and energy consumption. They will facilitate the development of parallel applications, reducing development and maintenance costs. Within this objective, on the other hand, computationally efficient solutions will be developed for applications that currently deal with open scientific challenges. The first is the efficient simulation of semiconductor devices that is required for fields as varied as the construction of last generation solar cells or the manufacture of new electronic devices. The solution of problems related to emergency landing situations of helicopters and drones, or the exploitation of resources associated with terrestrial LIDAR and photogrammetry information is also addressed. Finally, problems related to the recording of information from different sources, analysis of land cover, or the development of techniques that allow the mapping of plant resources using multi and hyperspectral images will be studied. All these applications work with large amounts of data and with efficient execution requirements and in many cases in real time. The results could be used to deal with the problem of forest fires.

In the second objective, HPC and Cloud for the processing of large volumes of data, solutions associated with Cloud computing are proposed, that allow a better use of the computational resources available in an organization for the efficient execution of Big Data tasks. These architectures will combine the use of dedicated clusters with on-demand clusters created with FPGA nodes and opportunistic computing systems. The availability of FPGAs that integrate general purpose processors with programmable logic in the same chip, will allow to obtain configurable systems with a general purpose processor and several processing units configuring this heterogeneous system. Security in the Cloud and container-based environments will also be analyzed and improvements will be proposed so that these deployments in public clouds are secure. Problems in the field of bioinformatics and semiconductor device simulation will be taken as applications for which these techniques will be tested.