Speaking from the kitchen of his California home this morning, NVIDIA CEO and Co-Founder Jensen Huang introduced the impressive NVIDIA A100. This is the first GPU based on the company’s Ampere GPU architecture, which, with more than 54 billion transistors, is the world’s largest 7-nanometer processor.
“The powerful trends of cloud computing and AI are driving a tectonic shift in data centre designs so that what was once a sea of CPU-only servers is now GPU-accelerated computing,” said Huang. “For the first time, scale-up and scale-out workloads can be accelerated on one platform. NVIDIA A100 will simultaneously boost throughput and drive down the cost of data centres.”
NVIDIA says the A100 represents the largest leap in performance across the company’s eight GPU generations — a boost of up to 20x over its predecessors — and that it will unify AI training and inference. The A100 is also built for data analytics, scientific computing and cloud graphics.
Huang never ceases to surprise with his amusing teases. Just two days before today’s keynote, a sneak peek video “What’s Jensen been cooking?” appeared on YouTube. Back in his kitchen, Huang says “I’ve got something to show you, it’s been cooking for a while…” He opens his oven and (with a bit of effort) removes a freshly baked motherboard, proudly proclaiming “it’s the world’s largest graphics card!”
A new technical feature called Multi-instance GPU will enable a single A100 to be partitioned into as many as seven separate GPUS depending on needs. The feature provides users the flexibility to optimize to scale of use and maximize investment return.
The US chip giant also announced the NVIDIA DGX A100 system, which combines eight A100 GPUs with the NVIDIA NVlink. To aid in the fight against COVID-19, NVIDIA said it has shipped one DGX A100 to the US Department of Energy’s Argonne National Laboratory, a multidisciplinary science and engineering research centre in Illinois.
On the commercial side, NVIDIA announced that companies such as Alibaba Cloud, AWS, Baidu Cloud, Google Cloud, Microsoft Azure, Oracle, and Tencent Cloud are planning to roll out A100-based services.
Journalist: Fangyu Cai | Editor: Michael Sarazen