Google launches A3 instance next month: with Nvidia H100, it can provide 26 exaFlops AI performance

CTOnews.com, August 31, Google Cloud announced the launch of an A3 virtual machine instance next month at today's Cloud Next conference. Google announced the A3 instance at the I / O developer conference held in May this year. The biggest highlight is that it is equipped with Nvidia H100 Tensor Core GPU to meet the needs of generative AI and large language models.

CTOnews.com previously reported that the A3 instance uses a fourth-generation Intel Xeon scalable processor, 2TB DDR5-4800 memory, and 8 Invidia H100 "Hopper" GPU, which achieves 3.6 TBps bisection bandwidth through NVLink 4.0 and NVSwitch.

The new A3 supercomputer "is designed to train and serve the most demanding artificial intelligence models that power today's generation of artificial intelligence and large-scale language model innovation". According to reports, this supercomputer can provide 26 exaFlops of artificial intelligence performance.

Google Cloud also introduced the new TPU v5e at today's event, which is by far the most cost-effective and accessible cloud TPU. These TPU or custom ASIC are designed to accelerate AI and ML workloads.

SDxCentral reported that the TPU v5e doubled its training performance per dollar and improved its reasoning performance by 2.5 times per dollar compared with the previous generation.

Related readings:

Google launches A3 Virtual Machines supercomputer: Nvidia H100 with 26 exaFlops AI performance