Reg readers have not one, but TWO teams in Folding@home top 1,000 as virus-bothering network hits 2.4 exa-FLOPS

Distributed computing project sails past anticipated raw power of El Capitan – and you folks are at the forefront

Give yourselves a pat on the back, ladies and gents – two teams made up of Register readers are in the top 1,000 out of more than 250,000 in the Folding@home distributed computing project for disease research.

The network has been used to simulate the folding of proteins implicated in a variety of illnesses for about 20 years, but at the end of February scientists on the project announced that they would be contributing resources to fight COVID-19, the potentially lethal respiratory infection caused by the new coronavirus that's rampaging across the globe right now and turning us all into bleary-eyed shut-ins.

The Register picked up the story three weeks ago, when Folding@home's raw computational power reached 470 peta-FLOPS, surpassing the world's most-powerful (publicly known) supercomputer – IBM's Summit at Oak Ridge National Laboratory.

It's not quite a like-for-like comparison: Folding@home is a massively distributed network, with volunteer computers on the ends of broadband internet links waiting for work, whereas supercomputers are tightly integrated for fast parallel processing with applications feeding in work from local storage via specialist networking – but at times like these, we all enjoy a bit of "good news".

It seems readers were all too happy to donate their spare CPU and/or GPU cycles to the cause. Commenters swiftly assembled the "Folding Vultures" team, named for The Reg's carrion-scoffing mascot. Not even a month later, the Folding Vultures are sitting at 883 out of 250,341 teams with 549 active CPUs within the last 50 days.

But it doesn't end there – another vulture-based Folding@home team was founded in 2007 to replace the group lost when a cancer-tackling distributed computing project through Grid.org closed. Ticking over quietly in the background these past 13 years, Vulture Central 2.2 holds an even more respectable position of 713. There's more about how points and rankings are calculated here and the top teams can be found here.

Obviously there's something compelling about harnessing home PCs to possibly make some progress towards a current threat, and the total computational power available to Folding@home has grown exponentially. A week after conquering Summit, the network went exascale, exceeding 1.5 exa-FLOPS, or 1,500,000,000,000,000,000 floating point operations per second. Where are we now? Take a look at the official stats – 2.4 EFLOPS.

With our collective power, we are now at ~2.4 exaFLOPS (faster than the top 500 supercomputers combined)! 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! pic.twitter.com/fdUaXOcdFJ

— Folding@home (@foldingathome) April 13, 2020

The virus research platform has gone, if you'll excuse the expression, viral.

To put this in perspective, Summit-dethroning El Capitan – the US Department of Energy supercomputer commissioned to simulate America's nuclear weapon stockpiles – is supposed to be capable of at least two EFLOPS on its own, and it's not even operational yet.

Meanwhile, Japan has tasked its newest supercomputer, Fujitsu's 400-PFLOPS Fugaku, with researching how new and existing drugs interact with the novel coronavirus in the hope of finding a suitable treatment. The US is also said to be doing similar through a consortium including Google, Microsoft, Amazon, HPE and, of course, IBM.

It should be said, however, that the calculations involved in simulating protein dynamics are enormous and these machines aren't magically going to cure COVID-19 overnight. The Folding@home team likens running a sim to buying a lottery ticket – the more people doing it, the more likely the project is to make a critical discovery. But it's not as though your spare cycles would be going to waste in any case. There's always important science to be done, and you can check out the many peer-reviewed papers made possible by the network here.

We can't all be "essential workers". We can't all volunteer with healthcare services or supermarkets. But I'd wager a good number of you are sitting on some beastly kit out there, and we can definitely all download a client and donate some of our surplus resources. It just goes to show what Reg readers can accomplish when we smash our heads together. ®