At the GTC 2021 Conference, Nvidia unveiled several innovative CPUs and GPUs.
The firm managed to introduce Grace, which is the first CPU made for the data center market specifically. The corporation presented the new RTX series card as well, which is intended for working professionals and manages to offer generational performance improvements, which are of high significance.
Nvidia became known as one of the best manufacturers when it comes to high-performance gaming GPUs. However, its contemporary output is considered to be a lot more diverse. It found tremendous success when it sold graphics processors which enable data centers to run a lot more efficiently throughout the previous few years. As a result of this, it derives 40% of the revenue from its server clients.
At GTC of 2021, Nvidia showcased that they had an even greater presence within the data center market through their Grace CPU.
The firm even designed the component that can tackle advanced applications such as AI supercomputing, natural language processing as well as recommender systems. When these are paired with its GPUs, the chip provides up to x10 more compute performance than contemporary systems. It also offers x30 the aggregate bandwidth of contemporary servers through 900GB per second through NVidia’s NVLink connection.
Aside from this, Grace even has an LPDDR5x memory subsystem that has twice the bandwidth alongside x10 times the energy efficiency when compared to LPDDR4 modules.
Nvidia also stated that their first CPU would be available for purchase in early 2023, and revealed that the U.S. Department of Energy’s Los Alamos National Laboratory, as well as the Swiss National Supercomputing Centre, would build new supercomputers which are powered by Grace to facilitate their research work. Nvidia is relatively new to the data center sector, and as such, its server GPUs have quickly become indispensable. Its components actually managed to play a critical role in the facilitation of the post-pandemic digitalization boom, and this means that it likely has insights into how large-scale HPC providers will need to evolve in order to address these requirements.
Due to this knowledge, the forthcoming chips could be compelling options for long-time Intel as well as AMD customers.
They also introduced the DGX Station 320G, which is a powerful workgroup appliance designed for data science teams to maximize their productivity as well as accelerate time to insights.
Introducing NVIDIA DGX Station 320G. Take a look inside this powerful new workgroup appliance, designed for #datascience teams to maximize productivity and accelerate time to insight. Learn more at #GTC21. https://t.co/14hDGYoNr5
— NVIDIA Data Center (@NVIDIADC) April 15, 2021
However, Nvidia is still focused on producing high-quality consumer GPUs as well.
It unveiled the new line of RTX series desktop as well as laptop GPUs that are specifically made using the Ampere architecture. These latest offerings are geared to artists, designers, and engineers who work remotely.
For desktop users, the firm managed to create the RTX A5000 and RTX A4000 Graphics processors, which offer double the AI processing as well as real-time graphical rendering speed of products from prior generations due to their upgraded RT, Tensor, and CUDA cores. Both of the products feature iterative improvements, and the A4000 comes with 16 GB GDDR6 SRAM, while the A5000 comes with 24 GB of GDDR6 SRAM.
In terms of the laptop lineup, they have the RTX A2000, RTX A3000, RTX4000, and RTX5000 GPUs.