Google Cloud has become the first cloud service provider to launch a cloud instance running Nvidia's new A100 GPU for accelerating artificial intelligence workloads.
Nvidia announced Tuesday that Google Cloud will make the A100 available through the company's new Accelerator-Optimized A2 virtual machine instance on the Google Compute Engine. The A100 will also soon make its way to Google Kubernetes Engine, Cloud AI Platform and other Google Cloud services.
The launch of Google Cloud's new A100-powered A2 instance comes roughly two months after the Santa Clara, Calif.-based chipmaker revealed the A100 and its 7-nanometer Ampere GPU architecture, which it said will "revolutionize" AI by unifying training and inference into one architecture that can outperform Nvidia's V100 and T4 GPUs several times over.
"Google Cloud customers often look to us to provide the latest hardware and software services to help them drive innovation on AI and scientific computing workloads," said Manish Sainani, director of product management at Google Cloud, in a statement. "With our new A2 VM family, we are proud to be the first major cloud provider to market NVIDIA A100 GPUs, just as we were with NVIDIA’s T4 GPUs. We are excited to see what our customers will do with these new capabilities."
The new A2 VM provides up to 16 A100s in a single instance, with a total of 640 GB of GPU memory and 1.3 TB of system memory. Thanks to Nvidia's NVSwitch interconnect, the 16 A100s can pool resources together to act as one giant GPU, but the individual A100 can also be partitioned into as many as seven separate GPU instances each because of the A100's multi-instance GPU feature.
Ian Buck, Nvidia's general manager and vice president of accelerated computing, said beyond AI training and inference, the A100 can also power data analytics, scientific computing, genomics, edge video analytics and 5G services, among other kinds of compute-intensive applications.
"Fast-growing, critical industries will be able to accelerate their discoveries with the breakthrough performance of A100 on Google Compute Engine," he said in a blog post. "From scaling-up AI training and scientific computing, to scaling-out inference applications, to enabling real-time conversational AI, A100 accelerates complex and unpredictable workloads of all sizes running in the cloud."
Before announcing the first A100-powered cloud instance, Nvidia said in June that more than 50 servers from OEMs like Dell Technologies, Cisco Systems and Hewlett Packard Enterprise will support the A100, thanks in part to a PCIe 4.0 card that fits in existing server motherboards. The A100's other form factor, the A100 SXM, is only support by Nvidia's HGX A100 server board.
In a recent interview with CRN, Buck said that the A100 will future-proof data centers for AI workloads and democratize AI, thanks to the GPU's unified training and inference capabilities that will pave the way for large-scale infrastructure investments,
"By having one infrastructure that can be both used for training at scale as well as inference for scale out at the same time, it not only protects the investment, but it makes it future-proof as things move around, as networks change — you can configure your data center in any way possible well after you've purchased and physically built it," he said in the interview.
Mike Trojecki, vice president of digital solutions and services at New York-based Nvidia partner Logicalis, recently said that Logicalis is looking at Nvidia as a key partner for driving sales for AI-based services and products this year and that the A100 will help unlock more opportunities than any previous GPU.
"The interesting thing here with the A100 is it brings AI down to a level that makes it available to everyone now," he told CRN in May, citing Nvidia's claim that its DGX A100 can perform the same level of training and inference work as 50 DGX-1 and 600 CPU systems at a tenth of the cost and a twentieth of the power. "It's not just this giant system that was out there where you couldn't afford it. This brings it down and makes AI consumable."