Power & Performance: Scaling Datacenters when the Chips are Down

Why it matters!?

Data volumes are growing rapidly—building opportunities for more insights, and better business outcomes. The capability of conventional datacenter processors, and the four walls of the datacenters they sit in—meanwhile—is not.

Scaling artificial intelligence can create a massive competitive advantage, and GPUs unlock those opportunities by delivering massive improvements in performance per watt.

But what about the workloads that aren’t yet accelerated? Will you be able to unlock the capacity to grow your revenue and support the increasing performance your existing customers demand?

Ian Finder, Principal Product Manager at NVIDIA Corporation shares his perspective on this dilemma exploring trends around CPU designs, Arm in the data center, and how architectures like NVIDIA’s Grace and Grace Hopper superchips are designed to unleash datacenter efficiency.

How do you plan for this and what are the most important trade-offs you need to consider? Ian shares practical use case examples to open up the group discussion on this critical challenge.