Nvidia and Microsoft accelerate AI processing on PCs

Nvidia and Microsoft declared Work to accelerate the AI ​​processing performance on the NVIDIA RTX based AI PC.

Generative AI converts the PC program to regulatory practices – digital people are written assistants, intelligent agents and creative vehicles.

NVidia RTX AI PCS strengthens this transformation with technology that allows you to practice this transformation with a simpler, generative AI and unlock more performance on Windows 11.

Tensortt for RTX AI PCS

TenSorrt, RTX has been reconstructed for EI PCs, only the engine building of the industry and the engine building on the device and 100 million RTX AI PCs combined with a package size of 8x for fast AI placement.

It was announced in Microsoft, TenSorrt for RTX, local, supported by Windows ML – a new inferencing stack that provides app developers with both an extensive device compatibility and art performance status.

The director of the product for the AI ​​PC in NVIDIA Gerardo Delgado, said the AI ​​PC’s NVIDIA RTX hardware, CuDA programming and AI models began in the press briefing. He noted that the high level, the AI ​​model is a set of mathematical operations along with a way to manage them. And the combination of transactions and how to use them, normal as a graphic in the learning of the machine.

He added that these operations are carried out with tensor cores. However, we are implementing them with a tensor core. the approach. “

First, NVIDIA must optimize the AI ​​model. It is forced to measure the size of the model so reduces the accuracy of the model or parts of some strata. After the optimized model of the NVidia, the Tensorrt consumes this optimized model of the optimized model, and then the NVidia mainly prepares a plan with pre-selection of kernels. “

If you compare this to Windows with a standard management of the AI ​​in the standard, NVIDIA can achieve a 1.6-time performance.

Now it will be a new version of Tensorrt for RTX to develop this experience. RTX is specially designed for AI PCS and provides the same tensortion performance, but instead of pre-forming Texier Engines in GPU, will be aimed at optimizing the model and send a general tensortion engine.

“Once the application is installed, TenSorrt for RTX will create a proper tension engine for a special GPU in just a few seconds. This develops the flow of work very much easier,” he said.

Among the results, the size of libraries, better performance for video generation and more quality Livestreams, Delgado said.

Nvidia SDKs, app creators simplify the combining AI features and accelerate their apps in GFORCE RTX GPU. This month, Autodesk, Bilibili, Chaos, LM Studio and Topaz releases updates to open RTX AI features and accelerations from Topaz.

NVIDIA NIM, which works in popular applications such as AI lovers and developers, NVIDIA NIM, pre-designed, optimized AI models, can easily start with optimized AI models. Flux.1-Schnell image generation model is already available as a NIM and has been updated to support the Popular Flowers.1-Dev NIM more RTX GPU.

RTX PC AI Assistant in the NVIDIA application, the Project G-Assist – NVIDIA application applied a simple way to build plug-ins. Now there are new community plug-insects including Google Gemini website search, Spotify, Twitch, IFTTT and SignalRGB.

Has the results of the AI ​​with Tensort for RTX

Today’s AI PC software provides a wide range of apparatus, but low performance or just performance or only certain equipment or model types require choices between frames that require more than one way.

The new Windows ML inference framework has been established to solve these difficulties. Windows ML OnNX is based on the time of working time and is provided by the manufacturer of any apparatus and the optimized AI executive of the stored. Windows ML for Geforce RTX GPUs automatically uses tenshes for RTX – uses a result library optimized for high performance and fast placement. Compared to DirecTML, TenSorrt presents 50% faster performance for AI business loads in PC.

Windows ML also offers the quality of life benefits for the developer. To manage each AI feature, you can automatically select the correct apparatus and download the executive provider for this device by eliminating the needs of these documents. This allows users to ensure the optimization of the latest texting performance to users as soon as they are ready for NVIDIA. Onnx works with any OnNX model Windows ML for being built on time.

To further enhance the experience for developers, the Tensorrt is reconstructed for RTX. To apply Tensort Engine to advance and apply them, TenSorrt uses the engine building on the device only in time to optimize the AI ​​model for RTX only for the user’s special RTX GPU. The library is corrected by reducing the file size through mass eight times. Tensortion for RTX, it will be available for developers through Windows ML before and will be available as a SDK that targets a June release directly in the NVIDIA developer.

Developers can learn more in Texorts for NVIDIA’s Microsoft Builder Blog, RTX launcher blog and Microsoft Windows ML Blog.

Expand the AI ​​ecosystem on Windows PC

The developers to add AI features or increase the performance of the application can hit a wide range of NVIDIA SDKs. These include COUDA and TENORTRT for GPU acceleration; DLS and optix for 3D graphics; Maxine for RTX Video and Multimedia; and Ace for Riva, Nemotron or Generative AI.

The best applications use this month to provide these month updates to provide NVIDIA unique features. Topaz releases a generative AI video model to improve video quality accelerated by CUDA. Chaos Enscape and Autodesk VRED are added to DLSS 4 for faster performance and better image quality. Bidibia integrates NVIDIA broadcasting features that allow Lividia to activate the NVIDIA virtual background within Lividia Livehime within Lvidia Virthime.

Local AI, NIM Microservices and AI plans

Starting AI in PC can be awesome. AI developers and lovers have to choose one of the 1.2 million AI models to embrace face-to-face, and measure the amount of a well-running format in the PC, to use it for a format to use and show it. Nvidia Nim, RTX GPUs facilitates a list of all files packaged in pre-packaged and optimized all files needed to get full performance. And like the container microservices, you can work without problems along the same NIM PC or cloud.

It is a NIM package – a predefined generative AI model with everything you need to manage it.

RTX is already optimized for Tensortion for GPUs, which is suitable for all the best AI applications used by API’s ApI’s API’s API today.

In Computex, NVIDIA, FLUX.1-Schnell releases NIM – Fixed image from black forest laboratories – Geforce RTX 50 and 40 series Flux.1-Dev NIM update to add compatibility for GPU. This NIMS adds to additional performance thanks to additional models, faster performance with Texort. In Blackwell GPUs, these are twice as many as the FP4 and RTX optimizations.

AI developers can also skip their work using NVIDIA AI Planeprints – NIM workflows and NIM.

Last month, NVIDIA has provided a strong way to manage the composition and camera angles of the images created using the 3D scene as a 3D scene. The developers can change the open source plan for the needs or expand with additional functionality.

The new project G-auxiliary plug-ins and sample projects are already available

NVIDIA was presented as a Assistant to G-Ash as an experimental AI assistant to the NVIDIA app recently. G-Assist allows users to manage the Geforce RTX system using simple sound and text commands by offering a more affordable interface compared to the handwriting of numerous inheritance control panels.

Developers can easily use the project G-ins, using plug-ins, test assistant work and publish via Nvidia’s discord and github.

To facilitate the start of creating plug-ins, NVIDIA has developed a Chatgpt-based application that allows you to develop code / low-coded development with natural language commands – NVIDIA, which is easy to use. These lightweight, community-managed supplements, direct JSON recipes and Python logic.

New open source samples are available in Github by demonstrating how your device’s PC and game can increase work streams.

● Gemini: Google’s existing twins using the cloud-based free LLM has been updated to include real-time search capabilities.

● IFTTT: IOTTT and home automation systems are compatible with hundreds of end points working with IFTTT, digital installation and physical environments.

● Discord: Easily share game points or share messages directly to the servers directly without breaking the gameplay.

Explore the GitHub Deposit for additional examples – Spotify, including silent music control with the status of Livestream, and more.

Project G-Assist – AI Assistant for RTX PC

Companies accept the EU as a new PC interface. For example, SignalRGB develops the G-Assist Plugin that allows single lighting control between many manufacturers. SignalRGB users will soon be able to install this plug directly from the SignalRGB application.

The project is invited to develop and get supported in the development of the NVidia developer, which is interested in developing and practicing with G-Assistant plugins, joining the Diskboard Channel, share creativity.

Each week, the RTX AI Garage Blog presents a Society managed AI updates and content for those who want to learn both NIM Microservices and AI plans, and AI plans, digital people, productivity applications and more.