NVIDIA announced its Xavier processor not too long ago, but now the company has officially launched its AI supercomputer chip - the first of its kind, ever.
The new Xavier board is a single-chip computer with over 7 billion transistors, with NVIDIA securing bragging rights of Xavier being capable of pushing over 20 trillion operations per second, at just 20W. Better yet, Xavier can be put together with a total of 50 boards, capable of driving up to 1 quadrillion operations per second, using 1kW of power - not too damn bad at all, if I might say so myself.
NVIDIA CEO and founder Jen-Hsun Huang sees a fast-approaching golden age of computing, with advances and leaps coming so quickly that it will pass our expectations. AI and supercomputing codes can be merged, and will see the offloading of traditional coding to technologies like neural networks - where we can expect things to get very, very interesting.
Huang said: "Deep learning is a supercomputing challenge and a supercomputing opportunity. Modern supercomputers should be designed as AI supercomputers. This means a system has to be good at computational science and data science and that requires an architecture that is good for both. We want to be able to support models that are very large and process may of those across multiple nodes, so interconnectivity is important. We have shown GPUs that can be shared in this way across massive GPU sets of nodes and the supercomputers of the future will be balanced by these two computational approaches with this architecture".