Altair > Case Studies > Enhancing Computational Science at Scuola Normale Superiore with PBS Professional

Enhancing Computational Science at Scuola Normale Superiore with PBS Professional

Technology Category

Infrastructure as a Service (IaaS) - Cloud Computing
Sensors - Temperature Sensors

Applicable Industries

Chemicals
Education

Applicable Functions

Facility Management
Product Research & Development

Use Cases

Inventory Management
Virtual Reality

Services

Hardware Design & Engineering Services

About The Customer

Scuola Normale Superiore (SNS) is a public institution of university education that was formally founded on October 18, 1810, by a Napoleonic decree. It is at the forefront of global research, with students selected only on the basis of merit. Lessons are held in seminar form, and an intimate link connects education and research. SNS research activities are traditionally divided into two primary areas: Arts and Sciences. Graduates include scientists, politicians, and intellectuals, many of whom have made remarkable achievements in their fields. The SNS Faculty of Science hosts several internationally recognized research groups in Biology, Chemistry, Physics, and Mathematics, which are involved in theoretical and applied research in nearly every scientific field.

The Challenge

Scuola Normale Superiore (SNS), a public institution of university education at the forefront of global research, has a mission to convert IT innovation into opportunities in computational science. The SNS Faculty of Science hosts several internationally recognized research groups in Biology, Chemistry, Physics and Mathematics, which are involved in theoretical and applied research in nearly every scientific field. To support state-of-the-art research investigations with a strong theoretical/computational background, SNS established DREAMSLab, a center dedicated to high-performance computing (HPC) and cutting-edge visualization and virtual reality techniques. The core of the DreamsHPC computational facility is a Dell cluster with over 2600 cores monitored in real time. However, given the complexity of the simulations running on the DreamsHPC cluster, it was necessary to carefully plan the software layer to be deployed on the infrastructure. The key requirements were overall ease of management of the already-present heterogeneous hardware, as well as significant simplification for the addition of new computing nodes in the existing structure and the distribution of computational programs on the nodes.

The Solution

For the HPC workload management, SNS chose Altair’s PBS Professional, an industry-leading product with thousands of users and support in 20 countries worldwide. PBS Professional was chosen for its efficiency, flexibility, and robustness. It allows for the allocation of available computational power in a self-consistent way by means of a queue-label for each node-family. With PBS Professional, the HPC users at SNS now have dynamically balanced access to the available computational resources to run their jobs in up to a month of wall-clock time. Moreover, the management of the submitted computational instances is completely transparent to the user, which is important because users want to follow the status of their simulations without losing time interacting with the OS environment. As a result of the PBS Professional installation, SNS has seen a significant increase in its occupation rate, with a 50% higher cluster utilization over the previous configurations.

Operational Impact

The implementation of PBS Professional has led to a significant increase in the occupation rate of the HPC cluster at SNS. This has resulted in a 50% higher cluster utilization over the previous configurations. PBS Professional has also improved the computational balance over the HPC cluster, avoiding potential conflicts among simulations that require a different ratio between computational intensity and memory bandwidth. Overall, the Scuola Normale Superiore has been able to optimize the returns on its HPC investments by employing PBS Professional. This has allowed users to increase cluster utilization while maximizing the performance of the entire infrastructure. The management of the submitted computational instances is completely transparent to the user, which is important because users want to follow the status of their simulations without losing time interacting with the OS environment.

Quantitative Benefit

The HPC cluster has significantly increased its occupation rate, with a 50% higher cluster utilization over the previous configurations.
PBS Professional has improved the computational balance over the HPC cluster, avoiding potential conflicts among simulations that require a different ratio between computational intensity and memory bandwidth.
The Scuola Normale Superiore has been able to optimize the returns on its HPC investments by employing PBS Professional, which allows users to increase cluster utilization while maximizing the performance of the entire infrastructure.