Supercomputing: An exascale-sized challenge?

[Michael_Bird]

Supercomputers are big, they're noisy, and they use more energy than a small town. They don't look like much from the outside, but the inside tells a different story! Supercomputers are helping us solve some of the world's biggest problems, and they could be coming soon to a desktop near you...

Today, host Michael Bird goes down the rabbit hole of high-performance computing. We talk parallel processing, exascale and the democratisation of HPC with Bill Mannel, HPE. Andrew Emerson PhD from Cineca tells us about how supercomputers led the charge to find therapeutic drugs against the coronavirus. Plus, AI and HPC Research Scientist Jacob Balma explains how the combination of machine learning and HPC is pushing boundaries that could change science, medicine and society for good.

Supercomputing shakedown: HPC, FLOPs, exsacale, and Moore's Law

Supercomputers often seem mysterious and impenetrable, and it's probably got something to do with the sheer amount of buzzwords and acronyms. HPE's Bill Mannel untangles parallel processing, FLOPs and exascale to help us understand what makes a supercomputer so super.

Although HPC has been the reserve of scientists and researchers for a long time, Bill says that's going to change. Data is booming, and new materials innovation are driving performance up and costs down. So, how long will be before organisations start using these incredible feats of technology for themselves?

Drug repurposing: supercomputers vs. COVID-19

When the coronavirus pandemic struck, scientists were under pressure to find therapeutic drugs at record speed. As part of the Exscalate4COV Consortium, researcher Andrew Emerson PhD at Cineca was able to run virtual experiments on billions of known drug molecules thanks to one of Italy's largest supercomputers. Andrew shares the results of this awesome critical compute project, and his thoughts on how HPC has changed medicine dramatically over the past few years.

HPC and machine learning: the future of personalised healthcare?

Machine learning requires big data sets to learn from, and big computers to learn on. So, what happens when you use a supercomputer to build a massive artificial neural network?

HPE's Jacob Balma gives us an overview of Phar.ML, a large-scale machine learning project that uses neural network to predict protein binding. We discover how supercomputers could soon help us model the human brain, and why the possibility of curing cancer and degenerative diseases is now within our sights.

Key takeaways:

• Exascale is not the limit. The more powerful we get, the more complicated problems we can solve. The obsession with power and speed isn't going anywhere soon!
• HPC is being democratised, with data booming, and materials innovation driving performance up and costs down, organisations might soon be using these incredible feats of technology for themselves.
• HPC and machine learning is the drug discovery of the future. Projects like Phar.ML demonstrate the incredible potential of combing these two disciplines.

Links and Resources:

Infographic: Everything You Need To Know About Supercomputers | US Department of Energy

How Parallel Processing Works | HowStuffWorks

High-Performance Computing | HPE

Aurora Supercomputer and Argonne National Laboratory | HPE & Intel

Cineca Supercomputer Centre, Italy

EXSCALATE4COV Consortium

Supercomputing and the Exascale Era | HPE

Supercomputers and Machine Learning: A Perfect Match | Inside BIGDATA

PharML.Bind COVID-19 Compound Affinity Prediction | Jacob Balma, HPE

PharML.Bind: Pharmacologic Machine Learning for Protein-Ligand Interactions | J.Balma, A.Vose

Michael Bird
Hewlett Packard Enterprise

twitter.com/HPE_UKI
linkedin.com/company/hewlett-packard-enterprise
hpe.com/solutions