NVIDIA announces Ampere GPU for AI and Data Centers

PUBLISHED BY
Anna Domanska



TAGS:


12 months ago




NVIDIA-DGX-A100-Inside-View-Ampere-GPU

NVIDIA DGX A100 – Inside View (Image: NVIDIA)

NVIDIA CEO Jensen Huang will deliver the GTC address on YouTube on May 14 and cover the company’s latest innovations in AI, data science, healthcare, and graphics. All companies talk of their achievements in their specialized field of expertise, but what has got the geeks in a tizzy is a teaser video that shows Huang showing a glimpse of the “world’s largest graphic card”.

The GTC 2020 Keynote address was to be delivered on March 23 at San Jose, California but was postponed due to the coronavirus pandemic. NVIDIA moved the event online. GTC Digital has attracted more than 45,000 registered users who have participated in 300 recorded talks. In the teaser video, CEO Huang says that there is something cooking in his kitchen, and then he pulls out the giant graphic card from the oven. Going by the way Huang is holding the GPU, it seems to be pretty heavy though NVIDIA has made no revelation about the GPU anywhere.

The internet has almost exploded with speculations from the geek population, with some believing that the new graphics card will be carrying the latest GPU architecture named Ampere. It is supposed to be a part of the GeForce RTX 3080 and GeForce RTX 3070 series.

The next-gen GPU is touted to be 75 percent faster than the current one and offers 50 percent more performance at half the power of Turing, the current GPU architecture.

The new Ampere family of GPUs will be GA103 and GA104. GA103 will reportedly be powering the GeForce RTX 3080, according to netizens who follow NVIDIA.

GA103 will reportedly pack 3840 stream processors, 60 SMs, and 10/20GB of GDDR6 on a 320-bit memory bus. The 10/20GB VRAM option is a first, according to tweaktown. The GA104 will have 3072 stream processors, 48 SMs, and 8/16GB GDDR6 on a 256-bit memory bus — this will take form in the GeForce RTX 3070, says the website.

Gamers are of the opinion that the general version of the GA103 card will have 10GB of GDDR6, while the pro levels cards in the Quadro RTX 3000 series will carry 20GB. NVIDIA might even offer 20GB of VRAM on a GeForce RTX 3080, but it is doubtful, say gamers.

The expected specs of the new GPU Ampere series:

GA103 (GeForce RTX 3080)

10/20GB GDDR6
320-bit memory interface
60 SMs
3480 stream processors
GA104 (GeForce RTX 3070)

8/16GB GDDR6
256-bit memory interface
48 SMs
3072 stream processors

According to initial news, the new GPU is expected to be built at the TSMC facility, the factory partner of NVIDIA. It will be built on a 7-nanometer manufacturing process. These Ampere-based GPUs will have thousands of cores dedicated to non-graphics tasks (CUDA cores), AI work (Tensor Cores), and real-time ray tracing (RT cores). AI 100 will be primarily used in data centers. The latest data center push is in response to the pandemic during which the need for cloud-computing has risen.

Update: CEO Huang said at a press meet, “My expectation is that Ampere is going to do remarkably well. It’s our best data center GPU ever made and it capitalizes on nearly a decade of our data center experience.”

Nvidia is boosting its Tensor cores to make them easier to use for developers, and the A100 will also include 19.5 teraflops of FP32 performance, 6,912 CUDA cores, 40GB of memory, and 1.6TB/s of memory bandwidth, revealed the company.

NVIDIA-DGX-SuperPOD-with-NVIDIA-DGX-A100-Ampere-GPU

NVIDIA DGX SuperPOD with NVIDIA DGX A100 (Image: NVIDIA)

Nvidia is combining these GPUs into a stacked AI system that will power its supercomputers in data centers around the world. Nvidia is also including 15TB of Gen4 NVMe internal storage to power AI training tasks. Researchers and scientists using the DGX A100 systems will be able to split workloads into up to 56 instances, spreading smaller tasks across the powerful GPUs.

“If you take a look at the way modern data centers are architected, the workloads they have to do are more diverse than ever,” explains Huang. “Our approach going forward is not to just focus on the server itself but to think about the entire data center as a computing unit. Going forward I believe the world is going to think about data centers as a computing unit and we’re going to be thinking about data center-scale computing. No longer just personal computers or servers, but we’re going to be operating on the data center scale.”

 

 

Whether you want to stay up-to-date on the latest business news, read in-depth CEO interviews, or find new ideas on leadership, management and innovation, Industry Leaders Magazine is here to suit your needs and help you stay more informed.

Avatar
Anna Domanska
Anna Domanska is an Industry Leaders Magazine author possessing wide-range of knowledge for Business News. She is an avid reader and writer of Business and CEO Magazines and a rigorous follower of Business Leaders.

Recent Posts

Australia’s Officeworks stops sales of Apple’s AirTags on child safety concerns

Australia’s Officeworks stops sales of Apple’s AirTags on child safety concerns

Officeworks, the Australian office supply store chain, has pulled Apple’s newly launched AirTags from its store following safety concerns for children from its button battery
9 hours ago
Ex Google AI scientist joins Apple after resigning in protest for unfair practices

Ex Google AI scientist joins Apple after resigning in protest for unfair practices

Apple, Inc. has hired a former Google AI scientist who resigned in protest against the firing of two employees from the Ethics division. Sammy Bengio, the ex-Google employee, will
2 days ago
Solid Power raises $130 million in second funding round from Ford and BMW

Solid Power raises $130 million in second funding round from Ford and BMW

Solid Power, a solid-state battery system startup, has raised $130 million in Series B funding led by Ford Motor Company and BMW Group. The Louisville, Colorado-based SSB developer
2 days ago
Intel plans to invest $3.5 billion in New Mexico chip unit

Intel plans to invest $3.5 billion in New Mexico chip unit

Intel plans to invest $3.5 billion in its New Mexico factory as part of a plan to expand domestic manufacturing investments. This is one of the largest single investments by the co
2 days ago
Credit Suisse loses $5.4 billion in Archegos Capital collapse

Credit Suisse loses $5.4 billion in Archegos Capital collapse

Archegos is a family investment firm run by fund manager Bill Hwang, who borrowed billions of dollars from nine global banks to speculate on stocks. The implosion of the fund cause
3 days ago
Barclays sees triple net profit in latest quarter

Barclays sees triple net profit in latest quarter

Barclays PLC logged a stellar performance in its first quarter with net profit almost tripling on the back of a brighter economic outlook for both the UK and US. The stimulation pa
4 days ago