T
he first machine to break the exaflop
barrier (that’s one quintillion
floating-point operations per
second) wasn’t a huge dedicated
supercomputer built by a large corporation
such as IBM, but a bunch of interconnected of
PCs with ordinary CPUs and gaming GPUs.
Welcome to the Folding@home project, which
is currently undergoing a massive boom in
popularity after it started targeting COVID-19
research. It’s effectively the world’s fastest
supercomputer, and your PC can be a part of it.
‘It’s estimated that we have more aggregate
compute power than the next top 100
supercomputers combined,’ says Dr Greg
Bowman, Director of Folding@home and
Associate Professor at Washington University
School of Medicine. ‘Before the COVID-19
work, we had about 30,000 active volunteers
on Folding@home, and something like 100
petaflops of compute horsepower. Since then,
we’ve had over 700,000 people download
the software, and there are at least a million
devices running Folding@home right now. Our
last estimate was that we’re at 1.5 exaflops of
compute power. It’s incredible.’What is Folding@home?
Folding@home is what’s known as a
distributed computing network, where
colossal computing jobs are broken up
into tiny chunks, which are then farmed
out to many machines over the Internet,
rather than trying to process them on one
massive local computer setup. It’s a concept
that’s been around for a long time – PC
veterans may well remember the (now
defunct) SETI@home project, which used
your spare CPU clock cycles to help look for
extra-terrestrial intelligence.
The Folding@home project, which first
started nearly 20 years ago, applies the same
principle to medical research. When the
project started, as its name suggests, its main
aim was to simulate how proteins fold. In the
most basic terms, a protein starts out as a long
string of amino acids, but it can’t perform its
biological function, such as an enzyme, untilBEN HARDWIDGE TALKS TO PROFESSOR GREG BOWMAN,
DIRECTOR OF FOLDINGHOME, ABOUT THE PROJECT’S
COVID19 RESEARCH, AND HOW YOU CAN GET INVOLVEDFEATURE/ ANALYSIS