Generative AI is rapidly ushering in a new era of computing for productivity, content creation, gaming, and more. Generative AI models and applications such as NVIDIA NeMo, DLSS 3 Frame Generation, Meta LLaMa, ChatGPT, Adobe Firefly, and Stable Diffusion use neural networks to identify patterns and structures in existing data to generate new, original content. generate.
Optimized for GeForce RTX and NVIDIA RTX GPUs, which deliver up to 1,400 Tensor TFLOPS for AI inference, generative AI models can run up to 5x faster than competitive devices. This is thanks to Tensor Cores (dedicated hardware in RTX GPUs built to accelerate AI computations) and regular software improvements. Enhancements introduced at last week’s Microsoft Build conference have doubled the performance of generative AI models such as stable diffusion that utilize new DirectML optimizations.
As AI inference on local devices grows, PCs will need powerful and efficient hardware to support these complex tasks. To meet this need, RTX GPUs add Max-Q low-power inference for AI workloads. GPUs run at a fraction of the power for light inference tasks and scale up to unmatched levels of performance for heavy generative AI workloads.
To create new AI applications, developers now have access to a full RTX-accelerated AI development stack running on Windows 11, making it easier to develop, train, and deploy advanced AI models. It starts with model development and fine-tuning using optimized deep learning frameworks available via the Windows Subsystem for Linux.
Developers can then seamlessly move to the cloud and train on the same NVIDIA AI stack available from all major cloud service providers. Developers can then use tools such as the new Microsoft Olive to optimize trained models for fast inference. And finally, AI-enabled applications and features can be deployed to his installed base of over 100 million RTX PCs and workstations optimized for AI.
said Pavan Davuluri, corporate vice president of Windows Silicon and Systems Integration at Microsoft. “By working with NVIDIA on hardware and software optimizations, we’re giving developers an innovative, high-performance, and easy-to-deploy experience.”
Over 400 RTX AI-accelerated apps and games have been released to date, with more to come.
In his opening keynote at COMPUTEX 2023, NVIDIA founder and CEO Jensen Huang introduced NVIDIA Avatar Cloud Engine (ACE) for Games, a new generative AI to power game development.
This custom AI model foundry service transforms games by bringing intelligence to non-playable characters through AI-powered natural language interaction. Middleware, tool and game developers can use ACE for Games to build customized voice, dialogue and animation AI models and deploy them in their software and games.
Generative AI on RTX Everywhere
From servers to clouds to devices, generative AI running on RTX GPUs is everywhere. NVIDIA’s Accelerated AI Computing is a low-latency, full-stack effort. Over the years, we have optimized every part of our hardware and software architecture for AI, including our 4th Generation Tensor Cores, dedicated AI hardware on RTX GPUs.
Regular driver optimization ensures the best performance. His latest NVIDIA drivers, combined with updates to Olive-optimized models and DirectML, bring significant speedups to developers on Windows 11. For example, for developers utilizing the DirectML optimization pass, stable diffusion performance improved by a factor of 2 compared to previous interference times. .
And with the latest generation of RTX laptops and mobile workstations built on the NVIDIA Ada Lovelace architecture, users can take generative AI anywhere. Our next-generation mobile platform brings new levels of performance and portability in a tiny 14-inch form factor and weighs nearly 3 pounds. Manufacturers like Dell, HP, Lenovo, and ASUS are leveraging RTX GPUs and Tensor Cores to power the generative AI era.
“As AI continues to be adopted across industries at an annual growth rate of over 37% through 2030, businesses and consumers will increasingly need the right technologies to develop and implement AI, including generative AI.” Lenovo has been developing products and solutions for AI workloads for many years and is uniquely positioned to power generative AI from devices to servers to the cloud.Our NVIDIA RTX GPU-powered PCs (select Lenovo ThinkPads, ThinkStation, ThinkBook, Yoga, Legion, LOQ devices, etc.) are enabling a transformative wave of generative AI that improves everyday user experience in saving time, creating content, getting work done, gaming and more.” — Daryl Cromer, Vice President and Chief Technology Officer, PCs and Smart Devices, Lenovo
“Generative AI is revolutionary and will be a catalyst for future innovation across industries. It gives developers confidence while driving a new era of generative AI.” — Jim Nottingham, Senior Vice President and General Manager, Z by HP
“Our recent work with NVIDIA on Project Helix is focused on making it easier for enterprises to build and deploy trusted generative AI on-premises. Bringing AI to the PC.Think of app developers trying to perfect their neural network algorithms while keeping training data and IP under local control.This is powered by NVIDIA RTX GPUs. And as the world leader in workstations, Dell is uniquely positioned to help users securely accelerate AI applications from the edge to the data center. — Ed Ward, President, Client Products Group, Dell Technologies
“The era of generative AI has arrived, requiring massive processing and fully optimized hardware and software. , and we look forward to seeing the AI revolution continue to take shape in ASUS and ROG laptops.” — Galip Fu, Director of Global Consumer Marketing, ASUS
Laptops and mobile workstations with RTX GPUs will soon be able to take advantage of the best of both worlds. AI inference-only workloads are optimized for Tensor Core performance while keeping GPU power consumption as low as possible, extending battery life and keeping the system cool and quiet. GPUs can dynamically scale up to maximize AI performance when your workload demands it.
Developers can also learn how to optimize their applications end-to-end and take full advantage of GPU acceleration via the NVIDIA AI Developer Site for Accelerating Applications.
