Google is about to launch A3 instances: equipped with NVIDIA H100, providing 26 exaFlops of AI performance

IT Home News on August 31, Google Cloud announced at the Cloud Next conference held today that A3 virtual machine instances will be launched next month. Google Cloud 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.

IT House has previously reported that the A3 instance uses the 4th generation Intel Xeon Scalable processor, 2TB DDR5-4800 memory, and 8 NVIDIA H100 "Hopper" GPUs, achieving 3.6 TBps through NVLink 4.0 and NVSwitch. Bisection bandwidth

The new A3 supercomputer is specifically designed to train and serve the most demanding tasks on the artificial intelligence models that drive today’s innovations in generative artificial intelligence and large language models. According to reports, this supercomputer can provide 26 exaFlops of artificial intelligence performance

At today’s launch, Google Cloud also introduced the new TPU v5e, which is the most cost-effective and accessible cloud TPU to date. These TPUs and custom ASICs are designed to accelerate artificial intelligence and machine learning workloads

According to SDxCentral reports, TPU v5e has doubled the training performance per dollar and 2.5 times improved the inference performance per dollar compared to the previous generation product

The above is the detailed content of Google is about to launch A3 instances: equipped with NVIDIA H100, providing 26 exaFlops of AI performance. For more information, please follow other related articles on the PHP Chinese website!