Abubakar Idris
What could be one of the world’s most powerful supercomputers is inching toward being completed and fully switched on after years of work.
Aurora, which is run by the Department of Energy, is about half the size of a football field, and, when operational, should be able to complete two quintillion calculations a second (for context, a quintillion is a billion billions, and most top-of-range smartphones can handle 10-15 trillion calculations per second).
First announced in 2015, Aurora is a next-generation supercomputer known as an “exascale” supercomputer — which is the name given to a machine that can perform one quintillion operations or more per second.
Aurora’s $600 million facility built by Intel and Hewlett Packard Enterprise (HPE) is housed at the Argonne National Laboratory and is powered by 60,000 sophisticated semiconductor chips, according to the Wall Street Journal. For context, Frontier, the world’s current top supercomputer housed at the Oak Ridge National Laboratory, has around 40,000 graphics processing units or chips.
When it is up and running in 2024, Aurora will be capable of performing cutting-edge science, from analyzing the brain to predicting the weather and discovering drugs faster than any other machine can now.
And thanks to its racks and racks of chips, Aurora will also have more memory than Frontier, allowing it to effortlessly handle more complicated operations and the biggest large language models ever to both process information and generate answers to challenging questions.
Aurora is already being tested for use across industries including climate research and the development of new batteries. Since batteries run on chemical reactions, the machine can run through billions of models to develop more efficient batteries using the right mix of components without requiring a single chemical test, said Venkat Srinivasan, a top battery scientist at the Energy Department’s Argonne lab.
No comments:
Post a Comment