RTX AI PC and workstations bring exclusive AI functions and peak performance to gamers, creators, developers and everyday PC users.
+++
In 2018, with the launch of RTX technology and the first consumer-grade GPU (GeForce RTX) specially built for AI, NVIDIA accelerated the transition to AI computing. Since then, AI based on RTX PC and workstation has flourished, and now it has become an ecosystem with more than 100 million users and 500 AI applications.
Nowadays, regardless of PC or cloud, generative AI is creating new possibilities. NVIDIA has been deeply involved in the AI field for many years, and has unquestionable professionalism, providing strong performance for all users and easily applying various AI functions.
Whether at home or at work, you can make full use of the power of RTX AI when using software to improve work efficiency and entertainment experience. Thanks to the AI features provided by GeForce RTX GPU-gamers can play favorite games at higher resolution and higher frame rate; Creators can focus on the idea itself without wasting time waiting for the system to load or repeating boring work; Developers can also use generative AI for prototype design and automated debugging, thus simplifying the workflow.
The field of AI is developing rapidly. With the deepening of research, AI will be able to handle more complex tasks. These stringent performance requirements will be met by RTX.
What is it? AI ?
In terms of its most basic definition, artificial intelligence (AI) is a more intelligent computing type. The ability of a computer program or machine to think, learn and take action without explicit coding with commands is AI.
AI can be regarded as the process of developing the following computer systems: systems for autonomously performing tasks, ingesting and analyzing massive data, and then identifying patterns in these data.
The huge and developing AI research field always focuses on the development of systems, which can perform tasks that originally required human wisdom at a speed beyond the ability of any individual or team. Therefore, AI is widely regarded as subversive and highly transformative.
The basic principle of AI system is to adjust itself when new information and new data are input into these systems, so as to actually learn previous experience or learn patterns from data. This self-learning ability enables the AI system to complete a wide variety of tasks, including image recognition, speech recognition, language translation, medical diagnosis, car navigation, image and video enhancement and hundreds of other use cases.
The next stage of AI evolution is content generation, that is, generative AI. It enables users to quickly create new content and iterate over it according to various inputs, including text, images, sounds, animations, 3D models or other types of data. Then, the generative AI will generate new content in the same form or in a new form.
Popular language applications such as ChatGPT based on cloud deployment allow users to generate long texts through short text requests. Image generators, such as Stable Diffusion, can convert descriptive text input into desired images. The new application can generate video from text and convert 2D images into 3D renderings.
GeForce RTX AI PC and NVIDIA RTX work station
AI PC is a computer equipped with special AI acceleration hardware. With or without dedicated acceleration hardware, the experience is very different. Without special hardware acceleration, we can only wait for 3D images to be loaded, and what you see is what you get through AI Denoiser.
On RTX GPU, these dedicated AI accelerators are called Tensor Core. Tensor Core can significantly accelerate AI performance when running extremely demanding work and entertainment applications.
One way to measure the performance of AI is in teraops. Terops is "teraops per second" (TOPS). Similar to the horsepower rating of the engine, TOPS can be used as a single indicator to let users know the AI performance of PC. This generation of GeForce RTX GPU can provide computing power, ranging from 200 AI TOPS to 1300 AI TOPS, including a variety of choices for notebook computers and desktop computers.
In contrast, at present, the computing power of AI PC without independent GPU is only 10 to 45 TOPS.
Faced with more and more types of AI applications, users will need a PC that can perform specific AI tasks locally (that is, on the device rather than in the cloud). The advantages of AI PC are: no need for networking, and computing power is at your fingertips; No need to wait, high response and low delay; No need to upload, safer and more assured.
For everyone. AI
RTX GPU not only brings strong performance, but also brings the unique functions of RTX technology. These AI functions are significant for all users.
From AI supersampling to intelligent and personalized chat bots, all types of users can benefit from AI.
RTX Video can use AI to oversample streaming video and display the video in HDR, thus converting SDR low-resolution video into vivid HDR video, with the resolution up to 4K. With one click, RTX users can enable this feature for almost all videos played in Chrome or Edge browsers once and for all.
NVIDIA Broadcast is a free application for RTX users. Its interface is simple and clear, and it has a series of AI functions that can improve the video conference and live broadcast experience. The application can eliminate unnecessary background sounds such as keyboard tapping, vacuum cleaner sound and children’s screams through noise cancellation and room echo cancellation functions. This application can also use the virtual background function to replace or blur the background through better edge detection technology. The application can also use the function of eliminating video interference to make the graphic outline in the low-quality camera image smoother. In addition, no matter where the user moves, the application can keep the user in the center of the screen through autofocus and eye contact functions, and make the user’s eyes look like they are looking directly at the camera.
Chat with RTX is an example of a personalized AI chat robot that can run locally. It is not only easy to use, but also can be downloaded for free.
The technical demonstration was released in January this year, and the content related to Google Gemma will be updated soon.
Users can easily connect local files on a PC to a supported large language model by simply putting the files into a single folder and indicating the location of the folder for Chat with RTX. After that, users can quickly query context-related content.
Chat with RTX runs on the local Windows of GeForce RTX PC and NVIDIA RTX workstation, so it runs very fast, and the user’s data does not need to be transmitted to other places. Chat with RTX does not rely on cloud-based services. Users can process sensitive data on local PCs, so there is no need to share data with third parties or connect to the Internet.
For gamers AI
In the past six years, the game performance has achieved a huge leap with the help of AI acceleration. Since 2019, gamers from all walks of life have been using NVIDIA DLSS to improve the frame rate and image quality. NVIDIA DLSS is a technology that uses AI to automatically generate pixels in the game. After continuous improvement, this technology can now increase the frame rate by up to four times.
With the introduction of light reconstruction function in the latest version of DLSS 3.5, the image quality of some popular games in the world has been further improved, setting a new standard for a more visual and immersive game experience.
At present, more than 500 games and applications have revolutionized the way users play and create through ray tracing, DLSS and AI empowerment technologies.
In addition to improving the frame rate, AI can also improve the way gamers interact with characters in the game and the way developers reproduce classic games.
With the help of NVIDIA ACE microservices (including voice and animation models empowered by generative AI), developers can add vivid intelligent digital avatars to games. ACE, which appeared at CES, won many awards, because it can bring the game characters to life and let people see the future of PC games.
NVIDIA RTX Remix is a platform for Modder, which can provide various generative AI tools to transform the basic textures of classic games into new 4K resolution materials based on physical properties. Modder can use this platform to create an amazing remake of RTX, a classic game. At present, several remake projects have been published or are under development, including "half-life 2" RTX version and "Portal)」RTX version.
Creator-oriented AI
AI can reduce or automatically perform tedious tasks, allowing creators to make time for pure creation, thus releasing their creative potential. The following functions run faster on a PC equipped with NVIDIA RTX or GeForce RTX GPU, and even some functions can only run on such devices.
The AI of Adobe Premiere Pro can enhance voice tools to eliminate unnecessary noise and improve the quality of conversation.
Adobe Premiere Pro’s "enhanced voice" tool is accelerated by RTX, which can use AI to eliminate unnecessary noise and improve the quality of dialogue clips, making them sound like professional recordings. This tool runs 4.5 times faster on RTX devices than on Mac. Auto refactoring is another function of Premiere. It can use GPU acceleration to identify and track more relevant elements in video, and intelligently reconstruct video content for different aspect ratios.
The magic mask function of DaVinci Resolve is another AI function that can save time for video editors. Previously, if editors needed to adjust the color/brightness of the subject in a shot or remove redundant objects, they had to use a combination of various rendering techniques or basic Power Window and masks to separate the subject from the background.
The magic mask function completely changed this workflow. With the help of magic mask, the user only needs to draw a line on the subject, and then the AI will process it for a moment, and then display the area that should be selected. In addition, GeForce RTX laptops run this function 2.5 times faster than the fastest non-RTX laptops.
There are many ways for AI to accelerate creativity. The above is just an example. At present, there are more than 125 AI applications accelerated by RTX.
Developer-oriented AI
With extensible environment, hardware and software optimization and new API,AI is improving the way developers build software applications.
NVIDIA AI Workbench can help developers quickly create, test and customize pre-trained generative AI models and LLM with PC-level performance and memory footprint. This is an easy-to-use unified toolkit, which can run on local RTX PC and can be extended to almost all data centers, public clouds or NVIDIA DGX clouds.
After building the AI model for PC use cases, developers can use NVIDIA TensorRT to optimize the model. This software can help developers make full use of Tensor Core in RTX GPU.
With Tensor RT-LLM for Windows, Tensor RT acceleration can now be used for text-based applications. TensorRT-LLM for Windows is an open source code base that can improve LLM performance, and contains pre-optimized model files suitable for popular models (including Google Gemma, Meta Llama 2, Mistral and Microsoft Phi-2).
Developers can also use TensorRT-LLM encapsulation for the OpenAI Chat API. With only one line of code modified, continue.dev (an open source autopilot program that can utilize LLM, suitable for VS Code and JetBrains) can use TensorRT-LLM on RTX PC to accelerate local LLM reasoning.
Every week, we will reveal the secrets of AI, make it easier for more people to master relevant technologies, and show RTX AI PC users brand-new hardware, software, tools and acceleration functions.
AI’s iPhone moment has arrived, and this is just the beginning.
###
Source: Internet
关于作者