High Performance Computing (HPC) - Computerphile

TL;DR

The University of Nottingham’s High Performance Computing facility, called Minerva, supplies the large-scale computing needed for research including fluid dynamics, aerospace, genomics, and astronomy. Its Linux-based nodes run hundreds of scheduled jobs at a time, typically for days, while the facility consumes substantial power and costs about £1.5–£2 million in capital. Read on to understand its hardware, workload, reliability, costs, and replacement plans.

Transcript

This this two-factor authentication; so basically to get in I need my card and then I need a PIN and this is a scrambler pad so basically every time that you look at that the numbers are in a different order. This is the High Performance Computing facility for the university of Nottingham. SEAN>> What do you use it for? All sorts of things it's bas... Read More

Key Insights

• ✋ The High Performance Computing facility at the University of Nottingham is used for research in various fields, highlighting its versatility.
• 🇨🇷 The cost of a high-performance computing system is significant, both in terms of initial capital costs and ongoing expenses like power and licensing.
• 👶 The facility aims to maximize the utilization of existing equipment while gradually replacing it with newer technology.
• ✋ Reliability is generally high, with individual nodes and disks occasionally failing.
• 😫 Setting up a high-performance computing system requires understanding the specific requirements of different research areas.
• ❓ The network connectivity, storage capacity, and parallel processing capabilities of the facility contribute to its efficiency.
• 🫢 The facility employs fire suppression systems using inert gas to protect the equipment without causing damage.

Questions & Answers

Q: What is the University of Nottingham’s High Performance Computing facility used for?

Students and researchers use it for calculations requiring large amounts of computing power. Examples include fluid dynamics, aerospace, genomics, astronomy, data manipulation, analytics, and visualization.

Q: What is the University of Nottingham’s HPC system called?

The university calls its main High Performance Computing facility Minerva. The facility shown is the third generation of HPC at the university.

Q: What operating system and scheduler does Minerva use?

Most of Minerva’s compute nodes run a version of Linux. PBS is used as the main scheduler, while the storage is described as fairly standard.

Q: How many jobs run on the HPC system, and how long do they take?

Hundreds of jobs may be running at any one time. Most run for days, although some can continue for months; jobs are not normally run for years.

Q: How much does a high-performance computing system like Minerva cost?

The estimated capital cost is about £1.5–£2 million, stated in the transcript as $2.1–$2.8 million. Ongoing expenses include thousands of pounds in power costs, as well as licensing and support.

Q: How much power and cooling does the HPC facility require?

The facility has about 250 kilowatts of air conditioning. When the particular block shown runs flat out, it draws about 70 kilowatts continuously.

Q: How reliable is the University of Nottingham’s HPC facility?

The facility is generally described as reliable, although individual compute nodes and disks can fail. The team reportedly experiences more problems with the air conditioning than with the computing equipment itself, and the system is shut down for maintenance once a year for about a day.

Q: How do you choose and purchase a high-performance computing system?

First, determine the mix of jobs the system must run because number crunching, data movement, analytics, and visualization require different hardware configurations. Then take those workload requirements and the available budget to a supplier and ask what system they can provide.

Summary & Key Takeaways

• The High Performance Computing facility at the University of Nottingham uses two-factor authentication for access, requiring a card and a PIN.

• The facility is used for research in fields such as fluid dynamics, aerospace, and genomics, requiring high computing capabilities.

• The facility has multiple racks of equipment, with blinking lights indicating data storage and network activity.