A new era of accelerated computing is set to begin with NVIDIA’s next-generation Blackwell GPU architecture. This will allow businesses to develop and execute real-time generative AI on trillion-parameter large language models. More details about NVIDIA’s Blackwell Architecture are below.
About Blackwell Architecture by NVIDIA
In comparison to the Hopper architecture, which it replaced, the Blackwell platform promises up to a 25-fold reduction in cost and energy consumption. The new GPU architecture, named for the trailblazing statistician and mathematician David Harold Blackwell, brings six game-changing innovations.
“Generative AI is the defining technology of our time. Blackwell is the engine to power this new industrial revolution,” said Jensen Huang, founder and CEO of NVIDIA. “Working with the most dynamic companies in the world, we will realise the promise of AI for every industry.”
Blackwell’s major innovations include the world’s most powerful chip (208 billion transistors), a second-generation Transformer Engine that doubles the support for compute and model sizes, a fifth-generation NVLink interconnect for facilitating fast multi-GPU communication, and sophisticated engines for data decompression, security, and dependability.
The brains behind Blackwell are the NVIDIA GB200 Grace Blackwell Superchip, which uses an incredibly fast 900GB/s NVLink interconnect to combine two B200 Tensor Core GPUs with a Grace CPU. Systems like the liquid-cooled GB200 NVL72 platform, which has up to 72 Blackwell GPUs and 36 Grace CPUs and provides 1.4 exaflops of AI performance, can be built using multiple GB200 Superchips.
NVIDIA has already gained support from major cloud providers such as Amazon Web Services, Google Cloud, Microsoft Azure, and Oracle Cloud Infrastructure to provide Blackwell-powered instances. Other partners in the hardware, software, and sovereign cloud space that are working on Blackwell products include Tesla, OpenAI, Microsoft, Meta, Dell Technologies, and many more.
Sundar Pichai, CEO of Alphabet and Google, said: “We are fortunate to have a longstanding partnership with NVIDIA and look forward to bringing the breakthrough capabilities of the Blackwell GPU to our cloud customers and teams across Google to accelerate future discoveries.”
The Blackwell architecture and accompanying software stack will enable new developments in a variety of fields, including engineering, chip design, scientific computing, and generative artificial intelligence.
“AI already powers everything from our large language models to our content recommendations, ads, and safety systems, and it is only going to get more important in the future,” said Mark Zuckerberg, the founder and CEO of Meta.
“We are excited to use NVIDIA’s Blackwell to support the training of our open-source Llama models and the development of the upcoming Meta AI and consumer products.”
Blackwell has the potential to be the driving force behind the eventual commercialization of real-time trillion-parameter AI, thanks to its enormous efficiency and performance gains.
