Friday, April 24, 2020

LIKE #SETI

Computer users donating spare processing power in the search for a coronavirus vaccine accidentally create the world's fastest supercomputer more powerful the then next 500 machines combined

Folding@Home is an app people install on their computers to donate resources

It allows thousands of machines to work together to solve complicated issues

It allows them to get a deeper understanding of how proteins work in the body

One major project is searching for how COVID-19 attaches itself to human cells


By RYAN MORRISON FOR MAILONLINE PUBLISHED: 17 April 2020

People running an app that uses part of their computer processing power to search for a coronavirus vaccine accidentally created the world's fastest supercomputer.

Known as Folding@Home, the technology uses thousands of computers around the world to work through large sets of numbers and complicated problems.

In the past month it has become so powerful that is has outpaced the 500 fastest supercomputers in the world combined in processing power.

There are a number of different COVID-19 projects from a range of universities and institutions making use of this giant distributed computer brain.

+3



The processing power is being used to understand how the spike from the COVID-19 virus - nicknamed Demogorgon after the monster from Stranger Things - enters the human cell

+3



The Folding@Home team say they've got so many users on the system at any one time their network is faster than any existing supercomputer

A number of studies are working to detect how the spike of the SARS-CoV-2 virus attaches itself to human cells and infect the body.


The virus is made of three proteins and they use a spike to grab on to a human cell that looks like the Demogorgon from Stranger Things.

It's how the virus penetrates the body and takes hold - blocking it is vital to developing future treatments and that's what Fold@Home is helping with.

Computer simulations powered by Folding@home are working to understand more about how spike proteins work.

'If you tried to simulate the opening of the spike on your home computer, you'd be lucky to see even part of the process within the next 100 years, said biochemist Greg Bowman from the Folding@Home team.

Since the outbreak of coronavirus that has most of the world in some degree of lockdown or isolation 700,000 new users have joined Folding@Home.

They've seen an increase of over 30,000 people running the app at any one time and it is produced a massive increase in computing power for the global system.

It now reaches 2.4 exaFLOPs of processing power - faster than the top 500 computers in the world combined.

The world's fastest single supercomputer is called Summit and is based at the US Department of Energy's Oak Ridge National Laboratory - it produces a peak performance of 187 petaFLOPs - just under 19 per cent of an exaFLOP.

'With our collective power, we are now at about 2.4 exaFLOPS (faster than the top 500 supercomputers combined),' Folding@Home tweeted.

'We complement supercomputers like IBM Summit, which runs short calculations using 1000s of GPUs at once, by spreading longer calculations around the world in smaller chunks.'


Folding@Home runs as an app on a computer and when the machine is idle uses the processor to crunch through data to find a cure for COVID-19 or in the past hunt for aliens

For a real world comparison - a top of the line MacBook Pro produces 153.6 gigaFLOPs and there are a million gigaFLOPs in a petaFLOP.

Folding@Home makes use of this remarkable power to split up complex protein models into tiny tasks that are distributed to thousands of computers.

The app lets you decide what percentage of your computer's processing power it can use and when it should run.

As so many people have installed the app purely to help in the coronavirus vaccine search, they have updated the software to let people prioritise COVID-19 projects.

Dr Bowman told the FT that Folding@Home has been used for everything from calculating how human hair grows to processes behind chemical reactions.

'Calculations are extremely computationally expensive and on a home PC they would take many years to complete,' he said in a video interview.

'We now get people to run chunks of simulations on their computers and that is spread around the world rather than on a single machine.'

No comments: