Pioneering distributed computing project [email protected] announced this month that it would no longer leverage the processing power of volunteers’ computers to search for alien life. But the [email protected] project which uses similar technology for medical research purposes is still going strong — stronger than ever, really.
[email protected] director Greg Bowman notes that the combined might of all the computers working together means that [email protected] now has “over 470 petaFLOPS of compute power,” making it more than twice as powerful as the world’s top supercomputer.
The uptick in performance comes as hundreds of thousands of computer users have dedicated spare CPU and GPU cycles to the project following [email protected]’s announcement that it’s working on projects to gain a better understanding of coronavirus/COVID-19.
As promised, here is our first glimpse of the #COVID19 spike protein (aka the demogorgon) in action, courtesy of @foldingathome . More to come! pic.twitter.com/iD2crCMHcX
— Greg Bowman (@drGregBowman) March 16, 2020
[email protected] uses simulations to better understand how proteins fold, which can lead to treatments for viruses, lcancer, and other conditions.
In other words, it’s unlikely that the work [email protected] is doing on the coronavirus will lead to a vaccine to help prevent the spread of COVID-19. But it could help researchers identify or develop drugs to help treat people who already have the virus, which could have an impact on just how deadly it is.
At this point, coronavirus research isn’t the only thing [email protected] is working on. But given the immediacy and widespread impact of the pandemic, Bowman says the project is “trying to bring the whole to bear on” COVID-19.
You can find instructions for installing the [email protected] client on your computer and dedicating unused resources to the analysis of COVID-19 at the project’s website. Once you’ve done that, the project may send small chunks of data to be processed on your computer when you’re not taking full advantage of your system’s resources for other activities. By combining the computing power of thousands of individual computers, the project now has the processing power of the world’s top 7 supercomputers combined.
Alternately, if you’re looking for other ways to help you could make a donation to [email protected] through Washington University in St. Louis, or contribute to any number of other national, regional, or local organizations that are doing their part to both combat the disease and to help protect the health and financial security of folks who have been affected.
Update: As of March 25th, 2020 the [email protected] project exceeded 1 exaflop (a quintillion operations per second) of performance. Â
via @drGregBowman and Tom’s Hardware
Intel foot dragging milking and withholding investment all these years is bad timing for this. Too little too late as usual. AMD to the rescue but volumes still not there. A side note it’d be interesting to see how much cpu vs gpu power this requires, and how software/os/kernel allocates each to such tasks as protein folding models.
Or could it be done better on FPGA or other exotic processor? Anyone studying this? Links?
This would be a funny thing for a time traveler to read if they came from a time period where they have determined P=NP, and they have a math formula for this.
Protein folding is a type of Constraint satisfaction problem (solving a sudoku is another example). The reason we need so much computing power is because we haven’t yet figured out a math formula to solve the problem in a reasonable amount of time. Solving it requires working through the whole data set.
That’s true, but we have developed a few algorithms to do calculations locally but not entirety.
The complexity increases exponentially as you zoom out, so it’s unlikely there will ever be a algorithm for it. That’s like what astro-physists are striving for in a unified equation. Too complex.