NVIDIA Boosts Generative AI with RTX GPUs, LLMs, Tools

NVIDIA utilizes cutting-edge tools, LLMs, and advanced RTX GPUs to power generative AI

NVIDIA has unveiled new AI laptops from leading manufacturers, new NVIDIA RTX-accelerated AI applications, and new tools for consumers and developers. The GeForce RTXTM SUPER desktop GPUs are designed to improve generative AI performance. NVIDIA is delivering tools to improve PC experiences with generative AI, building on its expertise in PCs. These tools include NVIDIA ACE microservices, NVIDIA TensorRT acceleration of the Stable Diffusion XL model, and NVIDIA RTX Remix with generative AI texture tools. DLSS 3 with Frame Generation technology will be used in more games.

Soon, the AI Workbench, a comprehensive toolbox for AI developers, will be accessible in beta. More pre-optimized models for PCs are now supported by NVIDIA TensorRT-LLM (TRT-LLM), an open-source library that quickens and enhances large language model (LLM) inference performance. Chat with RTX, an NVIDIA tech demo, is accelerated by TRT-LLM and enables AI aficionados to engage with their notes, papers, and other information.

The CEO and creator of NVIDIA, Jensen Huang, claimed that generative AI would revolutionize every sector, including gaming, and that it represents a major platform shift in computer history. Building on the more than 500 AI-enabled PC apps and games that NVIDIA RTX technology has previously boosted, NVIDIA is bringing breakthroughs to market across its whole technology stack.

RTX AI PCs and Workstations

The entire potential of generative AI on PCs is made possible by NVIDIA’s RTX GPUs, which can run a wide range of applications. of demanding applications, the Tensor Cores of these GPUs greatly improve AI performance. The GeForce RTX 4080 SUPER, 4070 Ti SUPER, and 4070 SUPER are among the recently released GeForce RTX 40 SUPER Series graphics cards that feature superior AI performance. AI video is generated 1.5 times quicker and pictures 1.7 times faster by the RTX 4080 SUPER GPU than by the RTX 3080 Ti GPU.

RTX AI laptops, which come with generative AI capabilities right out of the box, are being released by leading manufacturers. This month will see the release of these new systems, which will enhance performance by 20–60 times when compared to neural processing units. NVIDIA AI Enterprise software simplifies, secures, and advances data science and generative AI development on mobile workstations equipped with RTX GPUs. NVIDIA A800 40GB Active GPUs come with a three-year license for NVIDIA AI Enterprise.

New PC Developer Tools for Building AI Models

The AI Workbench, which NVIDIA unveiled recently, enables developers can use PC-class performance to design, test, and adapt pretrained generative AI models and LLMs. An easier way to access well-known repositories like Hugging Face, GitHub, and NVIDIA NGCTM is via the AI Workbench, which will soon be available in beta. Additionally, it provides an easy-to-use interface for collaboration, migration, and project replication.

Projects may be scaled to NVIDIA DGXTM Cloud, public clouds, or data centers, and then returned to local RTX systems for inference and customization. Through the integration of NVIDIA AI Foundation Models and Endpoints into the HP AI Studio, NVIDIA, and HP are streamlining the process of developing AI models. Users may now import, export, and use optimized models on cloud-based and PC platforms.

Developers may utilize NVIDIA TensorRT to improve AI models created for PC use cases, therefore using the Tensor Cores of RTX GPUs. With TensorRT-LLM for Windows, an open-source toolkit for speeding LLMs, NVIDIA has expanded TensorRT to text-based applications. Adding up to 5 times quicker performance over previous inference backends, the current release brings Phi-2 to the set of pre-optimized models for PC.

RTX-Accelerated Generative AI Powers New PC Experiences

New AI-powered PC apps are being released by NVIDIA and its partners at CES. NVIDIA RTX Remix is a platform that uses AI to remake vintage games, converting simple graphics into materials with 4K quality. Developers may incorporate dynamic, intelligent digital avatars into games with the help of NVIDIA ACE microservices. Up to 60% faster latent consistency and Stable Diffusion XL (SDXL) Turbo models are achieved using TensorRT.

The latest version of the Stable Diffusion WebUI TensorRT addon now supports LCM – Low-Rank Adaptation (LoRA), SDXL, and SDXL Turbo acceleration. Twelve of the fourteen newly revealed RTX titles will have NVIDIA DLSS 3 with Frame Generation, which quadruples frame rates above native rendering with artificial intelligence. AI aficionados may use retrieval-augmented generation (RAG) to link PC LLMs to their data via Chat with RTX, a tech demo that will be available later this month. The demo allows for rapid interaction with notes, papers, and other information and will be made available as an open-source reference project. TensorRT-LLM accelerates the demo.