CES 2022, Nvidia expands its tricks again! In addition to the much-anticipated new flagship graphics card 3090Ti, there is also the price-performance king 3050 for only US$249, and the most powerful notebook GPU in history—3080Ti. In addition, the Omniverse platform is also open for free .
Don’t wait, Huang didn’t come to this conference.
But Nvidia will finally live up to your waiting!
Last night, Nvidia officially announced that the “Metaverse” platform Omniverse was open for free, and it is not difficult to make Lao Huang’s kitchen.
Nvidia conference without graphics card is not complete!
No, RTX 3050, the king of GPUs with all-round price-performance ratio, is here. It costs only US$249. It can not only play 3A games at 60fps, but also supports DLSS.
In addition, RTX 3090 Ti’s new flagship graphics cards, RTX 3080 12GB, RTX 3070 Ti 16GB large memory graphics cards, and a new generation of game books that will be equipped with RTX 3080 Ti / 3070 Ti notebook graphics cards are all heavy eggs in this conference.
Oh, by the way, “Ti”, which has been controversial, has an official pronunciation!
That is-you can say “Ti” together or “T·I” separately.
Of course, when it comes to Nvidia, the first impression is that it’s a graphics card, although since the rise of the mining boom, we have not seen what Nvidia’s original graphics card looks like for a long time.
However, in terms of performance alone, Nvidia said it can continue to handle it, and the new flagship of the 30 series is here!
In Nvidia’s words, this is a “monster-level” GPU: GeForce RTX 3090 Ti.
It seems that when everyone is waiting for who will take over the Ampere architecture, Nvidia seems to think that the performance of the current 8nm GA102 chip has not been completely drained.
From the naming point of view, this new flagship is an upgraded version of the previous flagship card RTX 3090. In fact, the 3090 Ti has the same 24GB GDDR6X video memory as the 3090 in terms of video memory capacity, and the bandwidth is 21GB/s. The RTX 3090 is 19.5Gbps.
Nvidia said that the RTX 3090 Ti will also have 40 teraflops of GPU computing power, which is about 11% faster than the 36 teraflops of the RTX 3090. The light tracking power is 78 teraflops, and the AI task power is 320 teraflops.
For more detailed parameters of RTX 3090 Ti, NVIDIA did not mention it at the press conference. For example, it did not give the GPU base frequency and Boost frequency, but overall, the performance of RTX 3090 Ti is about 10% stronger than RTX 3090.
In addition, according to previous rumors, RTX 3090 Ti will be equipped with a “full-blood version” GA102 GPU, and the number of CUDA cores will reach an unprecedented 10,752.
The picture above shows the hardware parameter prediction of 3090 TI and comparison with 3090 by foreign media before the conference. It can be seen that from the data (video memory specifications, capacity, bus) that Nvidia has announced, the predictions in the left column of the table are accurate.
In terms of power consumption and the most critical price, NVIDIA kept a hand and did not announce it at this event, but said that “more details will be announced later this month.”
Foreign media theverge believes that from the 3090’s official starting price of $1,499, the official starting price of the 3090Ti may be around $2,000.
Of course, just look at the official price. What the actual price will be, the “waiting parties” have been aware of it for more than a year.
At present, due to the lack of available information, there is no actual game evaluation performance and power consumption and heat test on the 3090 TI.
According to previous predictions, the TDP of the 3090 TI single card may be 450w, which is 100w higher than that of the 3090. Without a power supply of 1000w or more, I am sorry for the title of “flagship”.
In short, the card has been released. As for whether it can be bought and whether it can be afforded, we still have to wait for “later this month” to see it again.
“Little Brother” 3050: Enjoy 3A masterpiece at 60 frames
Of course, the flagship card is indeed tall, but after all, not everyone has such high performance requirements. For ordinary PC users who do not have such high demand for games, NVIDIA will of course not give up.
This time NVIDIA released the “little brother” of the 30-series performance-level graphics cards: RTX 3050
According to previous release habits, the 3050 Ti, which was released together with the 3050, should have its “twin brother” 3050Ti, but this release seems to have broken this convention.
The GeForce RTX 3050 has a 128-bit bus width and is equipped with 8GB of GDDR6 video memory, with a video memory bandwidth of 224GB/s.
RTX 3050 is equipped with 2560 CUDA cores, 20 RT cores and 80 Tensor cores. The basic core frequency is 1.55GHz, and the Boost frequency is 1.8GHz. The thermal design power consumption (TGP) is 130W, and the public version design only needs a 4pin to supply power.
These parameter configurations are very similar to the previous GeForce RTX 3050 Ti predicted by the industry.
Nvidia GeForce RTX 3050 is priced at US$249 and is expected to go on sale on January 27.
Nvidia said that at 1080p resolution, the RTX 3050 can easily run current 3A game masterpieces at 60 frames. However, you may need to make some screen settings and turn off the light tracking effect.
To achieve this, a powerful CPU is needed. NVIDIA’s game test uses an Intel i9 processor.
Nevertheless, if you are still using a 1060 graphics card with Pascal architecture and have a limited budget, you can consider upgrading.
The strongest in history! Notebook 3080Ti>Desktop Titan
On the mobile platform, NVIDIA has also made big moves, announcing the launch of more than 160 laptops equipped with the new GeForce RTX 30 series GPU, using the new 4th generation Max-Q technology, which can further improve efficiency, performance and battery life.
Some notebook computers will be equipped with the new GeForce RTX 3080 Ti notebook GPU, which is the first time that the flagship 80 Ti GPU has appeared on the notebook platform.
The 3080 Ti notebook platform released this time is equipped with 16GB GDDR6 video memory, and its performance is even higher than that of the desktop TITAN RTX. The starting price is US$2499.
The GeForce RTX 3070 Ti notebook GPU can support a 100-frame/second display with a resolution of 1440 pixels, with a starting price of US$1499.
The laptop brand partners announced by NVIDIA this time include Alien, MSI and Razer, and games with these two graphics cards will be released on February 1.
The current 3A game masterpieces will put extremely high pressure on the performance of the CPU and GPU. When running under high voltage, the CPU and GPU often rob power resources. This problem may be even more prominent in game books with tight resources.
For 3A stand-alone masterpieces, the pressure on the GPU is often greater than that on the CPU. Nvidia launched the fourth-generation Max-Q platform to solve this problem.
Nvidia claims that Max-Q can control the CPU of a notebook computer and control the performance, power and temperature of the CPU as needed, thereby freeing up more resource space for the GPU and providing a better gaming experience.
It is actually a “CPU performance optimization tool”, but at present, NVIDIA has not announced the specific performance improvement of Max-Q.
Omniverse is free!
At the GTC conference in November last year, Nvidia released the paid Omniverse Enterprise Edition.
Today, “Engineer’s Metaverse” Omniverse will be open to artists and creators for free!
Download link: nvidia.com/omniverse
Nvidia said that Omniverse, which has invested hundreds of millions of dollars, is the culmination of its pioneering work for more than 20 years. It integrates graphics, artificial intelligence, simulation and scalable computing into a single platform to enhance existing 3D workflows.
Through Omniverse, GeForce RTX Studio creators can connect their favorite 3D design tools to a scene and simultaneously create and edit between applications.
Since the launch of the public beta a year ago, Omniverse has been downloaded by nearly 100,000 creators, who have used its core rendering, physics and artificial intelligence technologies to speed up the workflow.
Shanghai-based digital artist Zhelong Xu said: “With Omniverse, content creators get not only a fast renderer, but also a powerful platform with unlimited possibilities.”
Nvidia defines the Metaverse as an immersive and interconnected shared virtual world.
In these boundless worlds, artists can create unique digital scenes, architects can create beautiful buildings, and engineers can design new products for families.
Omniverse connects these independent 3D design worlds into a shared virtual scene.
In addition, in this special speech at CES, NVIDIA also announced the development of new platforms of Omniverse Machinima and Omniverse Audio2Face, new platform functions such as Nucleus Cloud and 3D market, and ecosystem updates.
Omniverse new features
1. Omniverse Nucleus Cloud
Artists can achieve “one-click collaboration” of large-scale Omniverse 3D scenes through the cloud, without the need to transfer large data sets. It’s like working on files shared in the cloud, but this time it’s a 3D scene.
2. A more complete Omniverse ecosystem
Creators will get more support in Omniverse’s ecology, and it will be easier to build the scenes they need.
Shutterstock, CGTrader, Sketchfab, and Twinbru’s TurboSquid have released thousands of materials for creators. These materials are based on the Universal Scenario Description (USD) format and can be found directly in the Omniverse launcher.
Afterwards, Reallusion’s ActorCore, Daz3D and e-on software’s PlantCatalog will also release their own Omniverse-ready materials soon.
3. Omniverse Machinima
For RTX creators who love games, characters, objects and environments from Mechwarrior 5 (Mechwarrior), Shadow Warrior 3 (Shadow Warrior), and Mount & Blade 2 (Mount & Blade) are all free to use .
Creators can use these materials to remix and create their own game movies by dragging and dropping these assets into their scenes.
4. Omniverse Audio2Face
Audio2Face is a revolutionary artificial intelligence application that can instantly animate 3D faces with just one audio track.
Now, “blendshape” support has been added, and the MetaHuman Creator application that can be directly output to Epic.
This leaves the tedious intermediate process to artificial intelligence, and artists and creators can spend more time on tasks that require creativity.
Canvas is upgraded again: new AI + resolution X4
The first generation of Canvas used GauGAN, a generative confrontation network composed of 5 million landscape photos.
Users only need to select various texture brushes, including hills, water, mountains or beaches, and then “paint” them loosely. After that, AI will take over and render these markers into realistic landscapes in real time.
However, if you look closely, the effect is still too fake.
This time, Canvas will usher in a big wave of upgrades: the resolution is increased by 4 times, and the generated images are more realistic, with higher definition, and fewer artifacts.
In fact, it is to replace AI with GauGAN 2.
On the basis of the first generation, GauGAN2 has received training on more than 10 million pictures, and has been able to convert natural language into landscape images.
Not much to say, let’s look at the effect first.
In addition, Nvidia has added five new materials to Canvas to create a richer landscape environment: straw, flowers, soil, clods and bushes.
These new materials can be further customized through nine preset styles. Of course, artists can also choose to upload their own images to create unique scenery.
With Nvidia’s Canvas, everyone can become an artist!
Posted by:CoinYuppie，Reprinted with attribution to:https://coinyuppie.com/lao-huang-didnt-come-nvidia-offered-rtx-3090ti-the-king-of-gpus-and-metaverse-platform-opened/ Coinyuppie is an open information publishing platform, all information provided is not related to the views and positions of coinyuppie, and does not constitute any investment and financial advice. Users are expected to carefully screen and prevent risks.