NVIDIA has unveiled groundbreaking AI performance optimizations and integrations for Windows, significantly enhancing the capabilities of NVIDIA GeForce RTX AI PCs and NVIDIA RTX workstations. These advancements, announced at Microsoft Build, promise to deliver up to 3x faster performance for large language models (LLMs) using ONNX Runtime (ORT) and DirectML with the new NVIDIA R555 Game Ready Driver. This development is a major leap forward for AI applications, enabling faster and more efficient execution of AI models locally on Windows PCs.
Enhanced Windows AI for Gamers, Creators, and Developers
- Up to 3x faster performance for LLMs with ONNX Runtime (ORT) and DirectML
- NVIDIA R555 Game Ready Driver
- Support for DQ-GEMM metacommand for INT4 weight-only quantization
- New RMSNorm normalization methods for Llama 2, Llama 3, Mistral, and Phi-3 models
- Group and multi-query attention mechanisms
- Sliding window attention for Mistral
- In-place KV updates for improved attention performance
- Support for GEMM of non-multiple-of-8 tensors
- WebNN acceleration with DirectML and ORT Web
- Up to 4x faster performance for models like Stable Diffusion, SD Turbo, and Whisper on WebNN
NVIDIA’s latest GPUs, equipped with dedicated AI acceleration, offer up to 1,300 trillion operations per second of AI performance. This power is harnessed across various applications, from boosting frame rates in games with NVIDIA DLSS to enhancing game characters with AI-driven dialogue and animation through NVIDIA ACE. Content creators benefit from AI-assisted production workflows in popular apps like Adobe Premiere and Blender, streamlining tasks such as 3D denoising and text-to-image generation. For developers, NVIDIA provides tools like NVIDIA AI Workbench and CUDA on Windows Subsystem for Linux, enabling the building and fine-tuning of AI models directly on their devices.
The new NVIDIA R555 Game Ready Driver, which includes these AI performance optimizations, is now available for download. This driver is compatible with a wide range of NVIDIA RTX GPUs, ensuring that both new and existing users can benefit from these enhancements. Additionally, Copilot+ PCs equipped with power-efficient systems-on-a-chip and RTX GPUs are set to be released in the coming months, offering increased performance for demanding AI workloads. Pricing for these new systems will vary based on configuration and specifications, but they are expected to be competitively priced to cater to gamers, creators, and developers alike.
For those interested in further exploring NVIDIA’s AI capabilities, the company offers a range of developer tools and frameworks such as NVIDIA TensorRT, NVIDIA Maxine, and RTX Video. These tools provide additional support for AI model development and optimization, making it easier for developers to create cutting-edge AI applications. Additionally, NVIDIA’s collaboration with Microsoft ensures continuous improvements and innovations in AI performance on Windows platforms, promising an exciting future for AI technology. Here are some other articles you may find of interest on the subject of Microsoft’s new Copilot PC range specifically designed for AI.
Latest trendsnapnews Gadgets Deals
Disclosure: Some of our articles include affiliate links. If you buy something through one of these links, trendsnapnews Gadgets may earn an affiliate commission. Learn about our Disclosure Policy.
