Nvidia Corp. is packing more powerful generative artificial intelligence processors into “compact workstations” with today’s launch of its new Nvidia RTX 2000 Ada Generation GPU, enabling users to run advanced AI applications on device.

The RTX 2000 is the most powerful graphics processing unit for workstations that the company has ever produced, offering up to 1.5 times the performance of its previous-generation RTX A2000 12GB. It’s designed to handle complex applications such as crafting 3D environments and refining industrial designs, and will pave the way for what Nvidia says will be an “AI-accelerated future.”

The company explains that modern multi-application workflows, such as generative AI, multi-display setups and high-resolution content put significant demands on GPU memory. To better handle this, the RTX 2000 Ada provides 16 gigabytes of on-board memory, ensuring it can support more realistic graphics, with photorealistic ray-tracing images rendered up to three times faster than previous-generation chips.

The chip is powered by Nvidia’s fourth-generation Tensor cores and provides up to 1.8-times greater AI throughput than the company’s older RTX A2000 12GB, together with a two-times boost in power efficiency. Other claims include a three-times performance increase for virtual reality workflows.

Nvidia said the RTX 2000 Ada Generation GPU can accelerate visualization and structural analysis workloads to enhance design precision in a number of industries. For instance, product designers and engineers using industrial PCs will be able to iterate more rapidly on new concepts with blazing-fast, photorealistic rendering and AI-powered generative design. Meanwhile, content creators will be able to edit high-resolution videos and images with much greater speed and precision, leveraging AI to create more realistic visual effects.

In addition, Nvidia envisages the RTX 2000 Ada Generation GPU powering real-time data processing at the edge, giving examples such as medical devices and manufacturing equipment, and AI-powered intelligence in retail environments.

Nvidia rolled out a number of early adopters of its latest GPU platform, including the 3D product design software maker Dassault Systèmes SE. According to Dassault’s graphics applications research and development director Olivier Zegdoun, the new chip is especially beneficial for users of its SOLIDWORKS design software. “It delivers excellent performance for designers and engineers to accelerate the development of innovative product experiences with full-model fidelity, even with larger datasets,” he promised.

Rob Wolkers, owner and senior industrial design engineer at Rob Wolkers Design and Engineering, emphasized the chip’s ability to provide more powerful compute and horsepower in industrial design scenarios. “Equipped with next-generation architecture and a large frame buffer, the RTX 2000 Ada Generation GPU improves productivity in my everyday industrial design and engineering workflows, allowing me to work with large datasets in full fidelity and generate renders with more lighting and reflection scenarios 3x faster,” he said.

Nvidia said the RTX 2000 Ada Generation GPU is being launched alongside the release of the latest RTX Enterprise Driver, which includes a range of features designed to enhance graphics workflows. For instance, Video TrueHDR is a new tone-mapping capability for standard dynamic range and high dynamic range video that delivers an expanded color range and higher brightness for content viewed in web browsers such as Chrome and Edge. The driver also adds support for Video Super Resolution and TrueHDR, enabling low-quality video to be enhanced and upscaled to HDR.

The driver also includes TensorRT-LLM, an open-source library that optimizes and accelerates inference performance for large language models, plus the ability to offload certain tasks from the central processing unit to the GPU via the execute indirect extension application programming interface.

Nvidia said the RTX 200 Ada is available now via its global distribution partners, and will be integrated into workstations from Dell Technologies Inc., HP Inc. and Lenovo Group Ltd. in April.

Images: Nvidia