Alphabet’s Google has recently claimed that its AI supercomputers are faster and more power-efficient than comparable systems from Nvidia. Google uses its own custom chip called the Tensor Processing Unit (TPU) for more than 90 percent of the company’s work on artificial intelligence training. The Google TPU is now in its fourth generation, and Google has strung more than 4,000 of the chips together into a supercomputer using its own custom-developed optical switches to help connect individual machines.
Improving connections has become a key point of competition among companies that build AI supercomputers. Google’s PaLM model, its largest publicly disclosed language model to date, was trained by splitting it across two of the 4,000-chip supercomputers over 50 days. Google’s supercomputers make it easy to reconfigure connections between chips on the fly, helping avoid problems and tweak for performance gains.
Google’s supercomputer has been online inside the company since 2020 in a data center in Mayes County, Oklahoma. Midjourney used the system to train its model, which generates fresh images after being fed a few words of text. For comparably sized systems, Google’s supercomputer is up to 1.7 times faster and 1.9 times more power-efficient than a system based on Nvidia’s A100 chip.
Google did not compare its fourth generation to Nvidia’s current flagship H100 chip because the H100 came to the market after Google’s chip and is made with newer technology. However, Google hinted that it might be working on a new TPU that would compete with the Nvidia H100 but provided no details. The paper published by Google detailing its supercomputer was scientific.
The flexibility of Google’s supercomputers allows them to change the topology of the supercomputer interconnect to accelerate the performance of an ML (machine learning) model. Google has a healthy pipeline of future chips.
For the latest tech news, follow Techbar Nepal on Facebook, and Instagram.
For Digital Products and Services: Maurisys Software.
