March 17 - News broke that NASA (the National Aeronautics and Space Administration), which once led the world, hasn't been doing so well in recent years. Many major space projects are not only seriously over budget but also way behind schedule.
Now, NASA has finally found the "culprit": the agency's supercomputers are too outdated.
NASA Supercomputing
NASA currently operates five supercomputers housed at two facilities: the NASA Advanced Supercomputing (NAS) facility in Ames, California, and the NASA Center for Climate Simulation (NCCS) in Goddard, Maryland.
The most powerful of these is Aitken, with a performance of just 13.12 PFlops (1.312 quadrillion floating-point operations per second), which is used for NASA's Artemis program to return to the Moon.
The others are Electra (8.32 PFlops), Discover (8.1 PFlops), Pleiades (7.09 PFlops), and Endeavour (15.48 TFlops).
Not only is the performance of these supercomputers underwhelming, but their architectures are also technologically outdated, still relying almost entirely on traditional CPU processors.
Despite having a total of over 18,000 GPUs at its disposal, NAS has only deployed 48 GPUs, and NCCS has even fewer.
NASA's own report acknowledges that its supercomputing infrastructure is severely inadequate, significantly impacting NASA's mission and suffering from poor management, inefficient utilization, and various security vulnerabilities, strongly recommending a shift to modern, GPU-based supercomputing.
Bottom line: fund it.
