At its I/O developer conference, Google today announced a public preview of a full cluster of Google Cloud’s Cloud TPU v4 Pods. Google unveiled the fourth iteration of its Tensor processing unit at last year’s I/O conference, and a TPU Pod consists of 4,096 of these chips. Each chip has a peak performance of 275 teraflops, and each pod promises 1.1 exaflops of combined computing power.
Google now operates a full cluster of eight of these pods in its Oklahoma data center with a peak aggregate performance of 9 exaflops. Google believes it is the world’s largest publicly available ML hub in terms of cumulative computing power while running on 90% carbon-free energy.
Those clusters are provided by ML (machine learning) capable supercomputers (meaning they are well suited for ML workloads such as NLP, recommendation models, etc. These supercomputers are powered by ML hardware – such as GPUs (graphics processing units) and are Built on CPU and memory. With 9 exaflops, we believe we have the largest publicly available ML cluster.
At I/O 2021, Google CEO Sundar Pichai said the company will soon have “dozens of TPU v4 Pods coming online in our data centers, many of which will be at or near 90 percent. Running on carbon-free energy. And our TPUv4 Pods will be available to our cloud customers later this year.” Obviously, that’s a bit longer than planned, but given the context that we’re in a global shortage of chips, these are custom chips after all.
Ahead of today’s launch, Google worked with researchers to expose them to these clusters. The researchers reported satisfaction with the performance and scalability provided by TPU v4 with its fast interconnect and optimized software stack, and liked the ability to set up their own interactive development environment with the new TPU VM architecture, as well as use their preferred Flexibility of frameworks including JAX, PyTorch or TensorFlow,
Google says users will be able to slice and dice the new cloud TPU v4 cluster and its pods to suit their needs, whether it’s access to four chips (which is the bare minimum for a TPU virtual machine) or thousands of chips (but also Not too many, as there are only so many chips available).
As of now, these clusters are only available in Oklahoma. “We conducted an extensive analysis of various locations and determined that Oklahoma, with its exceptional carbon-free energy supply, is the best place to host such a cluster. Our clients can access it from virtually anywhere, ‘ a spokesman explained.