Search

In biology, dynamical models are used to understand and predict the behavior of biochemical systems composed of numerous species and reactions. Model parameters represent aspects of the studied system and, therefore, modifications of model parameters correspond to perturbations of the real system. Several methods of model analysis are based on model perturbation by means of its parameter values. Thus, the execution of parameter sweep applications (PSAs) over large parameter spaces becomes very CPU intensive and time consuming.

The INFN-CNAF data centre hosts the Italian Tier~1 site for the Worldwide LHC Computing Grid (WLCG), while also serving several other research and technological transfer programs. The challenges posed by the upcoming runs of LHC, together with the opportunity of moving the data centre itself to a bigger site, require a thorough redesign of its monitoring system. The large but heterogeneous amount of logging data and metrics produced daily are fundamental for monitoring activities and, once harmonised, can also be used to build Predictive Maintenance models based on Big Data techniques. In this work we describe the Big Data Platform, a new monitoring infrastructure under development at CNAF. The Big Data Platform relies on a modular, highly scalable architecture based on open source technologies and able to exploit modern frameworks such as containerisation and cloud support. It is capable of collecting data from heterogeneous data sources, clean and harmonise them, and store them as JSON files on different solutions, based on the needs of the end user. Data can then be visualised using Kibana, or analysed through a platform based on Jupyter Notebooks.

The concept of digital twins has growing more and more interest not only in the academic field but also among industrial environments thanks to the fact that the Internet of Things has enabled its cost-effective implementation. Digital twins (or digital models) refer to a virtual representation of a physical product or process that integrate data from various sources such as data APIs, historical data, embedded sensors and open data, giving to the manufacturers an unprecedented view into how their products are performing. The EU-funded IoTwins project plans to build testbeds for digital twins in order to run real-time computation as close to the data origin as possible (e.g., IoT Gateway or Edge nodes), and whilst batch-wise tasks such as Big Data analytics and Machine Learning model training are advised to run on the Cloud, where computing resources are abundant. In this paper, the basic concepts of the IoTwins project, its reference architecture, functionalities and components have been presented and discussed.

In this work a previously published bioinformatics pipeline was reimplemented across various computational platforms, and the performances of its steps evaluated. The tested environments were: (I) dedicated bioinformatics-specific server (II) low-power single node (III) HPC single node (IV) virtual machine. The pipeline was tested on a use case of the analysis of a single patient to assess single-use performances, using the same configuration of the pipeline to be able to perform meaningful comparison and search the optimal environment/hybrid system configuration for biomedical analysis. Performances were evaluated in terms of execution wall time, memory usage and energy consumption per patient. Our results show that, albeit slower, low power single nodes are comparable with other environments for most of the steps, but with an energy consumption two to four times lower. These results indicate that these environments are viable candidates for bioinformatics clusters where long term efficiency is a factor.

The embedded and high-performance computing (HPC) sectors, that in the past were completely separated, are now somehow converging under the pressure of two driving forces: the release of less power consuming server processors and the increased performance of the new low power Systems-on-Chip (SoCs) developed to meet the requirements of the demanding mobile market. This convergence allows the porting to low power embedded architectures of applications that were originally confined to traditional HPC systems. In this paper, we present our experience of porting the Filtered Back-projection Algorithm to a low power, low cost system-on-chip, the NVIDIA Tegra K1, which is based on a quad core ARM CPU and on a NVIDIA Kepler GPU. This Filtered Back-projection Algorithm is heavily used in 3D Tomography reconstruction software. The porting has been done exploiting various programming languages (i.e. OpenMP, CUDA) and multiple versions of the application have been developed to exploit both the SoC CPU and GPU. The performances have been measured in terms of 2D slices (of a 3D volume) reconstructed per time unit and per energy unit. The results obtained with all the developed versions are reported and compared with those obtained on a typical x86 HPC node accelerated with a recent NVIDIA GPU. The best performances are achieved combining the OpenMP version and the CUDA version of the algorithm. In particular, we discovered that only three Jetson TK1 boards, equipped with Giga Ethernet interconnections, allow to reconstruct as many images per time unit as a traditional server, using one order of magnitude less energy. The results of this work can be applied for instance to the construction of an energy-efficient computing system of a portable tomographic apparatus.