The previous six months have observed a good deal of unparalleled activity throughout the world, most of it bad. A person vivid spot in the gloom — however of class, a rather conditional one — has been the unexpected explosion in computational electrical power getting flung at analyzing and comprehension COVID-19.
It wasn’t that extensive back we have been reporting on the Folding@Dwelling network’s document-breaking 1.5 exaFLOPS of general performance. Now, the process has broken 2.5 exaflops. At that pace, it’s reportedly more rapidly than the sustained general performance of all Leading500 supercomputers combined.
With our collective electrical power, we are now at ~2.4 exaFLOPS (more rapidly than the prime 500 supercomputers combined)! We complement supercomputers like IBM Summit, which operates short calculations applying 1000s of GPUs at at the time, by spreading for a longer time calculations close to the entire world in smaller sized chunks! pic.twitter.com/fdUaXOcdFJ
— Folding@dwelling (@foldingathome) April 13, 2020
There is a quite major asterisk to this, having said that. Folding@Dwelling is still ramping up the sum of perform they can assign to the community. If you really do not come to feel like this problem has improved a great deal in current months, it’s simply because the new equipment have been signing on as quickly as extra perform can be farmed out.
On April 13, the staff declared that Folding@Dwelling would now exam simulations of how existing antiviral prescription drugs like Remdesivir may well interfere with the ordinary replication of COVID-19 inside the physique. The goal for the task is to consider all of the Fda-accepted NTP (Nucleotide TriPhosphate) prescription drugs now on the current market to see which may well be the most productive at killing coronavirus. The task is also constantly doing the job to distribute a lot more perform models to all readily available devices, so if you have discovered your process getting idle recently, you have got two options: Hold out for some of the more recent perform models coming down the line, or transfer to a task like Rosetta@Dwelling, which is commonly smaller sized and could also likely use some aid.
I’m genuinely curious to see what arrives of all the computer time dedicated to resolving COVID-19. We have under no circumstances used this a great deal horsepower attacking a single virus in these kinds of a concentrated way, and what we study here could be very important to extensive-time period viral exploration on this and other initiatives. In no way in human historical past — not with the human genome, or AIDS, or SARS, or even most cancers exploration — has so a great deal horsepower been so unilaterally concentrated on a single goal. I really do not know if it’ll pay out dividends, but if it does, we’ll have realized anything about just how a great deal compute horsepower it requires to travel these varieties of breakthroughs.