Finland’s CSC Chronicles the COVID Research Performed on Its ‘Puhti’ Supercomputer – HPCwire

CSC, Finlands IT Center for Science, is home to a variety of computing resources, including the 1.7 petaflops Puhti supercomputer. The 682-node, Intel Cascade Lake-powered system, which places about halfway down the Top500 list, has managed to make major cultural and scientific waves over the last year thanks to its extensive use for COVID-19 research. In a new review article, CSC is highlighting the wide variety of coronavirus research hosted by Puhti through the course of the pandemic.

In March, CSC like many supercomputing centers around the world announced that it would be fast-tracking COVID-19 research. In CSCs case, approved projects were directed specifically to its Puhti system, with the center initially allocating a third of the supercomputer to the fast track for COVID-19.

In total, CSC awarded 15 applicant projects access to the Puhti fast track. The projects occupied much of the system from the spring through the middle of summer. However, CSC notes, the final fast track load on the system did not occupy that full third of the supercomputer for the duration of the pandemic: rather, it consumed a seemingly paltry 5.42 percent of Puhtis total usage over the course of 2020.

CSC and Puhtis breakthrough moment in COVID-19 research came early in the pandemic with bombshell simulations of viral particle spread (pictured in the header) that showed that a cough could transmit infectious COVID particles at considerable levels as far as 13 feet away, where they would linger for a number of minutes. These simulations one of which was presented in terms of a grocery store aisle captured the worlds attention at a time when the extent of COVIDs airborne transmission was not yet fully known or feared.

But this was far from the only application for Puhti, which is also involved in the global hunt for therapeutics and vaccines. One University of Helsinki researcher, for instance, is using Puhti to study cotransins, small molecules that are sometimes able to obstruct SARS-CoV-2 as it attempts to infect human cells; another group used Puhti to combine molecular dynamics and machine learning and gain better insights into the main protease of the virus. Yet others explored aspects of COVID-19 drug development ranging from the spike proteins and ACE2 receptors to drug repurposing and protein-protein inhibitors.

Other researchers used Puhti to investigate variants and mutations of the virus, which are of increasing concern as vaccinations promise to stamp out current forms of the virus in a number of hotspot countries. CSC reports that virtually all of the thousands of coronavirus genomes detected in Finnish patients were classified using Puhti, with new samples regularly arriving.

With the need for the fast track dwindling (CSC also introduced its 5.39 Linpack petaflops Mahti supercomputer last summer, diminishing competition for time on Mahti), the center is phasing out its COVID fast track this month. Mahti will itself be dwarfed by the 375 Linpack petaflops LUMI, the pre-exascale EuroHPC supercomputer that is slated to begin operations this year.

To learn more about the COVID research hosted by Puhti over the last year, visit CSCs roundup article here.

See the original post:

Finland's CSC Chronicles the COVID Research Performed on Its 'Puhti' Supercomputer - HPCwire