by
John R. Fischer, Senior Reporter | August 21, 2023
Harvard researchers have used cloud-computing as an alternative source to supercomputers for simulating a medical application for treating heart disease.
Doing heavy duty medical research that requires a supercomputer, but don't have access to one? Use the cloud to clone one.
In a study evaluating a new therapy for dissolving blood clots and tumor cells in the human circulatory system to treat heart disease, professor Petros Koumoutsakos, of the Harvard John A. Paulson School of Engineering and Applied Sciences (SEAS), overcame this limitation by using Google’s cloud platform to clone a supercomputer.
Working alongside Google, Citadel Securities, and researchers from ETH Zurich in Switzerland, Koumoutsakos and his team applied extensively tuned code to the platform and found in initial tests that it was 80% as efficient as a supercomputer. They have added one state-of-the-art code and an open-source commodity code (LAMMPS) for particle simulations for GPU- and CPU-based architectures.
“We hope to show that easy access to massively available cloud computing resources can significantly reduce time to solution, improve testing capabilities, and reduce research costs for some of humanity's most pressing problems,” said Koumoutsakos in a statement.
But while researchers and technology companies are increasingly looking to adopt cloud-based technologies, the platforms are primarily designed to perform small, computing tasks, such as video streaming, rather than meeting their scientific demands.
Bill Magro, chief high-performance computing technologist at Google Cloud,
told Reuters that bringing cloud computing to the same level as supercomputers requires changing the software, networking, and physical design of cloud-based infrastructure.
"Folks are realizing the potential for cloud to solve problems and technical scientific engineering computing to really unlock productivity and get to better answers, better insights, faster," he said.
