Nvidia presents Ampere GPU architecture to boost AI and the first target of coronavirus

Ampere will eventually replace Nvidia's Turing and Volta chips with a single platform that simplifies alignment of the Nvidia GPU, Huang said in a previous briefing with members of the media on Wednesday. While consumers are widely aware of Nvidia for its video game hardware, the first versions of Ampere are targeting artificial intelligence needs in the cloud and for research.

"This is undoubtedly the first time that we have unified the acceleration workload of the entire data center on a single platform," said Huang.

Nvidia discovered years ago that its gaming hardware is beneficial for machine learning thanks to its parallel processing design: when researchers try to teach algorithms with data, GPUs help generate more data at a faster rate. Since then, it has constantly developed products based on these needs in high performance computing, data centers and autonomous driving.

See also: Wall Street sees many benefits from these oppressed tech stocks

A 7 nanometer processor containing more than 54 billion transistors, Ampere takes the idea of ​​parallel processing and multiplies it: each individual A100 GPU, the first released with Ampere, can be divided to execute up to seven actions or engage in One Need, Huang said. The company has consolidated eight of these GPUs into the DGX A100, which can handle up to 56 tasks at a time, or be combined into one large task, and achieve up to 5 petaflops of AI performance.

"Because it is fungible, you don't have to buy all these different types and configurations of servers," said Huang.

The first DGX A100 systems, starting at $ 200,000, were delivered to the Argonne National Laboratory outside of Chicago earlier this month. Researchers at the Department of Energy at Argonne will use the power of AI to study COVID-19 and possible remedies for the disease.

"The computing power of the new DGX A100 systems arriving in Argonne will help researchers explore treatments and vaccines and study the spread of the virus, allowing scientists to do years of AI-accelerated work in months or more. in days, "Rick Stevens said the director of the Laboratory of Computer Science, Environment and Life Sciences at Argonne said in a press release.

Nvidia has also detailed other approaches to COVID-19 by researchers using Nvidia's machine learning capabilities, including sequencing the coronavirus genome that causes COVID-19 in just seven hours and detecting billion possible drug combinations in a day. Nvidia also took stock of its Clara platform for AI in healthcare, adding AI models that can be used to detect and study infected patients using chest scanner data.

Opinion: Nvidia Has Become An Energy Broker For The Next Wave Of Data Center Technology

Many researchers cannot afford a complete DGX A100 system, so cloud providers plan to buy A100 GPUs and sell remote access. Nvidia said 18 cloud providers are planning to integrate A100 chips, including some of the largest in the United States and China, Amazon.com Inc.
AMZN
+ 0.46%

AWS, Microsoft Corp.
MSFT
-1.51%

Azure, Alphabet Inc.
GOOGL,
-1.95%
GOOG,
-1.92%

Google Cloud, as well as cloud offerings from Alibaba Group Holding Ltd.
VASE
-0.42%

, Baidu Inc.
BIDU
-2.03%

and Tencent Holdings Ltd.
TCEHY
+ 4.49%

.

"I hope the Nvidia A100 is in the clouds," Huang said naturally during the briefing on Wednesday.

Nvidia did not publish information on mainstream GPUs using Ampère on Wednesday, but when a reporter asked about the difference between Ampère's mainstream and commercial approaches, Huang said: "There is a huge overlap in architecture, but not the configuration."

Nvidia also announced products intended to be used "on the periphery", which means the calculation of sensor data. For example, Nvidia said its EGX A100 offering could manage hundreds of cameras scattered around an airport, while the smaller EGX Jetson Xavier NX would take a collection of cameras to retail outlets.

Read: Intel and Nvidia make Merger Monday a very successful sequel

The company's efforts in the automotive sector have led to new alliances, including collaboration with the BMW Group on robotic equipment to help build cars. "The BMW Group's use of Nvidia's robotic platform Isaac to reinvent its factory is revolutionary," said Huang in a press release.

Nvidia was worth between $ 5 billion and $ 20 billion for almost the entire period between the dot-com crash and 2016, but its progress in machine learning has led to much higher valuation in recent years. Shares closed at a record high of $ 322.62 on Monday as the company neared a $ 200 billion market capitalization for the first time, closing Wednesday's session with one more valuation of $ 191 billion. So far this year, stocks have gained 32.3% despite the COVID-19 pandemic, while the S&P 500 index
SPX
-1.74%