How Machine Learning Is Changing Data Centers

Machine learning training involves tremendous data-crunching

Machine Learning And The Changing Face Of Today’s Data Centers

By Priti Shetti

Machine learning holds a good future but it requires huge computational power. The powerful GPUs of Big Sur helped Facebook crunch significantly more data, which resulted in a dramatic reduction of time to train its neural networks.

IBM’s Watson supercomputer is known to process at the rate of 80 teraflops. To imitate a high-functioning human’s capability to answer questions, Watson uses 90 servers. These servers are combined with a data store of 200 million plus pages and six million logic rules. The applications of Watson’s cognitive computing technology are endless.