Nvidia unveiled several innovative new electronic components at its GTC 2021 Conference.
Most notably, the firm introduced Grace, its first CPU made specifically for the data center market.
The corporation presented new RTX series graphics cards intended for working professionals that offer significant generational performance improvements.
Nvidia’s First Server-Class CPUs
Nvidia rose to prominence as a high-performance gaming GPUs manufacturer, but its contemporary output is considerably more diverse. It found tremendous success selling graphics processors that enable data centers to run more efficiently in recent years. As a result, it currently derives around 40 percent of its revenue from its server clients.
At this year’s GTC, Nvidia signaled its intent to become an even greater presence within the data center market with its Grace CPU.
The firm designed the component to tackle highly advanced applications like AI supercomputing, natural language processing, and recommender systems. When paired with its GPUs, the chip provides 10 times the compute performance of contemporary systems. It also offers 30 times the aggregate bandwidth of contemporary servers via a 900GB per second via Nvidia’s NVLink connection.
In addition, Grace features an LPDDR5x memory subsystem that has twice the bandwidth and 10 times the energy efficiency of LPDDR4 modules.
Nvidia stated its first server CPU would be available for purchase in early 2023. It also revealed the U.S. Department of Energy’s Los Alamos National Laboratory and the Swiss National Supercomputing Centre would build new supercomputers powered by Grace to facilitate their research work.
Although Nvidia is relatively new to the data center sector, its server GPUs have quickly become indispensable. Its components played a critical role in facilitating the post-pandemic digitalization boom. That means it likely has insights into how large-scale HPC providers need to evolve to address the “new normal.”
Because of that knowledge, its forthcoming chips could be a compelling option for long-time Intel and AMD business customers.
New RTX GPUs for Remote Work Professionals
Despite its recent focus on expanding its corporate portfolio, Nvidia is still keen to produce high-quality consumer hardware.
To that end, it unveiled a line of RTX series desktop and laptop GPUs made using its Ampere architecture. However, opposed to its hard-to-find gaming products, its latest offerings are geared toward artists, designers, and engineers working remotely.
For desktop users, the firm created the RTX A5000 and RTX A4000 graphics processors. They offer double the AI processing and real-time graphical rendering speed of products from the prior generation thanks to their upgraded RT, Tensor, and CUDA cores. Both products feature significant iterative improvements; the A4000 comes with 16GB GDDR6 SRAM while the A5000 houses 24GB of GDDR6 SRAM.
Nvidia’s new professional laptop lineup includes the RTX A2000, NVIDIA RTX A3000, RTX A4000, and RTX A5000 GPUs. The components possess “next-generation RTX technology,” which means they provide ray tracing and enhanced AI and compute acceleration. The product family supports Nvidia Omniverse, a collaborative real-time simulation platform based on Pixar’s production pipeline.
The chipmaker’s new laptop GPUs are also equipped with Max-Q system optimization technology and up to 16GB of memory.
Nvidia’s desktop products will be available worldwide later this month. It intends to launch its revamped laptop graphics processors in new notebooks made by its OEMs partners later this quarter.
Based on its GTC 2021 product introductions, Nvidia is primed to cement its position as one of the world’s foremost semiconductor manufacturers.