The training of these models can take days, weeks, and even months, depending on complexity of the model and the type – and amount – of data being processed.Not all data centres are equipped to handle the rapidly emerging, compute-intensive applications used in AI and machine learning. To achieve the best performance for these types of AI applications, organisations need data centres with an optimised design, infrastructure, and facilities and, crucially, the right power sources to support them.An example of this is DeepL, a technology company specialising in natural language translation. DeepL was founded with the goal of adding fuller context and a better understanding of local vernacular to machine translation. As such, the company needed the right power, compute and infrastructure to drive its 5.1 petaflop supercomputer, which has enough power to translate one million words in less than a second and serves as the translator’s brain. Supercomputers like DeepL’s are energy-intensive and require specialised high-performance computing (HPC) infrastructure and support to enable them to operate fluently.
Read the full article in the September issue of PBSI
Print this page | E-mail this page
UL standards presentation: The opportunity in North America
Download a copy of our digital magazine