Supercomputers

GTC Spring announcements aim for AI computing at the top and bottom of the spectrum.

Tightly integrated with Nvidia GPUs, the company’s first CPU is meant for the near future when AI models reach trillions of parameters.

Proliferation of AI, in combination with cloud platforms making it easier to test the waters, have led more IT organizations to turn to HPC-style infrastructure.

Although HPC systems traditionally come with a high CapEx, clouds now offer a cheaper initial on-ramp – and their ability to handle demanding AI/AL workloads makes them an attractive option.

Researchers “have open questions that they just can’t get answered with the compute resources that they have available to them.”

The Fugaku cluster has more than 150,000 processors and roughly 2.8 times the performance of the second-fastest supercomputer, US DoE's Summit.