Home News Nvidia Unveils A100 GPU and Ampere Architecture at GTC 2020 Keynote, to...

Nvidia Unveils A100 GPU and Ampere Architecture at GTC 2020 Keynote, to be a Revolution for AI

Nvidia GTC 2020 yet another Keynote to be shifted to an Online-only event amidst the pandemic

Nvidia planned to hold a proper GPU Technology Conference 2020 keynote had to be canceled due to the Coronavirus outbreak. Thus, the company decided to unveil its products via a pre-recorded keynote presentation yesterday. The company’s drive towards development and releasing new products has unabatedly continued as NVIDIA’s next-generation GPU architecture – Ampere, and the first products that will be using it were announced.

Nvidia GPU
Image Courtesy: Nvidia

The company was expected to unveil enterprise-related products along with its latest GPU architecture, and Nvidia didn’t disappoint. Ampere is the name of Nvidia’s 7nm architecture for its GPUs and compute units. While we don’t have a lot of details about the new architecture itself, there sure were a bunch of enterprise products that were announced, which use the Ampere architecture.

A100 GPU, HGX A100, and DGX A100

Nvidia’s A100 GPU its first GPU using the new Ampere architecture. The GPU packs in it 54 billion transistors and Nvidia calls it “the world’s largest 7-nanometer processor”. The A100 GPU also comes with Nvidia’s third-generation of Tensor Cores, which Nvidia claims offers 20 times faster AI performance. It also, for the first time, supports double-precision performance. The company is claiming a 2.5 times performance boost over the Tesla V100.

Alongside the A100 GPU, Nvidia also announced hardware that uses these A100 GPUs in multiple combinations. The DGX A100 announced packs in eight A100 GPUs that off a whopping 5 petaflops of AI performance. Together, the eight A100 GPUs in the DGX A100 offers 320GB GPU memory and 12.4 TBps bandwidth.

The customizable HGX A100 baseboard was also announced by Nvidia which can be used by OEMs to create their own A100 based server systems. Nvidia partnered with Dell, HP Enterprise, Gigabyte, and SuperMicro among others to offer HGX A100 based systems to enterprise customers.

Lastly, the company also announced the EGX A100 edge compute unit. This can “carry out real-time processing and protection of the massive amounts of streaming data from edge sensors“. The new Mellanox SmartNIC was also announced by the company which is present inside the EGX A100. As we expected, these are all enterprise-grade announcements. Another aspect to note is that Ampere based graphic cards are also in the works. We don’t have dates yet, but we do expect Nvidia to talk more about consumer GPU’s in upcoming months. Its about time for the rumored RTX 3000 series graphics cards to be revealed.


Please enter your comment!
Please enter your name here