Gamingdeputy reported on March 17 that although NASA has been using the world's most cutting-edge technology and has made many major discoveries in human history, an internal NASA Office of Inspector General General's special report revealed that some project schedules were delayed because NASA's supercomputer capabilities were insufficient to meet mission requirements.
Advertisement
The report pointed out thatNASA's supercomputers still rely mainly on CPU central processing unitssuch as one of its flagship supercomputers,It has 18,000 CPUs but only 48 GPUs.. You know, GPUs can provide far more computing power than CPUs for parallel processing tasks that are very common in scientific simulations and modeling.
Currently, NASA has five high-end computing (HEC) units at the NASA Advanced Supercomputing Center (NAS) at Ames Research Center (Ames) in California and the NASA Center for Climate Modeling (NCCS) at Goddard Space Center in Maryland. equipment. These equipment include Aitken, which supports the “Artemis Program” to return to the moon and establish a long-term base, the powerful Electra, and the Discover, which is used for climate and weather simulations. ), Pleiades and Endeavor for climate simulation, astrophysics research and aerospace modeling.
Gamingdeputy noticed that these machines almost entirely rely on old CPU cores. For example, all NAS supercomputers use more than 18,000 CPUs and only 48 GPUs, and NCCS has even fewer GPUs.
Advertisement
The report quoted NASA high-end computing officials as saying that there are many reasons why the system cannot be modernized, including supply chain issues, the need for modern programming languages, and a shortage of technical personnel needed to implement new technologies. The report states that the inability to upgrade existing high-end computing infrastructure will directly impact NASA's ability to achieve its exploration, science and research goals.
The audit also found that NASA's high-end computing operations were not centrally managed, resulting in inefficiencies and a lack of consistency in usage policies between internal and cloud computing resources. This uncertainty leads to hesitancy to use cloud computing resources due to unknown scheduling methods or assumed higher costs. Some missions are forced to acquire their own infrastructure to avoid waiting to use prime supercomputing resources that are overloaded in large part by not employing the latest high-performance computing technology.
Additionally, the audit found that security controls on high-end computing infrastructure were often bypassed or not implemented, increasing the risk of cyberattacks.
The report concludes by recommending that moving to GPUs and modernizing code is key to meeting NASA's current and future needs. This will greatly enhance its computing capabilities in fields such as scientific simulation and modeling, and help it successfully complete various space exploration and scientific research missions.