In August 2019, Cerebras debuted the Wafer Scale Engine (WSE) – the first AI chip to have 1.2 trillion transistors (the most ever), 18 gigabytes of on-chip memory (the biggest of any chip on the market today), and 400,000 processing cores (again, the fastest of all commercially available today).
But in just three months, Cerebras has outdone itself, with the launching of CS-1, complete AI solution built with its own WSE, cooling, networking, and storage system. It also comes with its own equipment and software required to operate and integrate the processor into the data center. CS-1 claims that it can deliver the performance of 1,000 of the best GPUs combined. It is easy to deploy and use, and it only uses 20 kilowatts of power.
The first CS-1 computer is currently being used by the US Department of Energy’s Argonne National Library. It is used in neural networks in cancer research, studies about black holes, and treatment of brain injuries. To learn more about the Cerebras chip, read the following article written by Dean Takahashi for Venture Beat.
Read Original Article
Click the button below if you wish to read the article on the website where it was originally published.
Click the button below if you wish to read the original article offline.