release date, price, technical sheet, all the info

With its graphics cards based on the Ada Lovelace architecture, Nvidia is taking a new step. Its GeForce RTX 4000 promise exceptional performance, even more oriented towards ray tracing, and featuring the new DLSS 3. The RTX 4080 and RTX 4090 have already been announced and should soon be joined by the RTX 4070 Ti, the old RTX 4080 12 Go.

The generation of Nvidia GeForce RTX 3000 graphics cards will gradually give way to the new Nvidia GeForce RTX 4000, the first models of which are official. These GPUs are based on a new Ada Lovelace architecture, incorporate new technologies and have even more impressive technical specifications than their predecessors. Discover with us everything you need to know about the RTX 40 family.

To read :

What graphics cards are available?

As usual, Nvidia is not launching all of its GPUs at the same time, and is starting with the high-end segment. The manufacturer has officially introduced three models of graphics cards for the GeForce RTX 4000 series: the RTX 4090, the RTX 4080 16 GB and the RTX 4080 12 GB. But the brand has been heavily criticized for the nomenclature it has chosen, from many consumers do not understand why the RTX 4080 16 GB and RTX 4080 12 GB, which have very different characteristics and performance, have the same name. Nvidia then changed tack and decided not to launch the RTX 4080 12 GB on the market.

To date, the series therefore only has two references. The Nvidia GeForce RTX 4090 has been available since October 12, 2022. As for the Nvidia GeForce RTX 4080, it has been on the market since November 16, 2022. The RTX 4000 generation should experience less shortage concerns over the long term than that of RTX 3000, for which it was almost impossible to find a graphics card at its recommended price. The excitement around cryptocurrency mining has waned along with the value of cryptocurrencies and rising energy costs. In addition, production lines are starting to return to a more regular pace after the pandemic, despite lingering episodes of confinement in China. But we must still expect difficulties in taming the launch of GPUs, especially with regard to the GeForce RTX 4080.

According to rumors, the RTX 4070 Ti would be presented on January 3, 2023. The first tests would be published the next day, January 4, for orders to open on January 5. But these dates have not yet been confirmed by Nvidia.

In addition, we also know that Nvidia is preparing an RTX 4060 as powerful and less expensive than an RTX 3070. In any case, this is what a Lenovo executive revealed. This entry-level GPU would therefore be 20% more powerful than the RTX 3060, while its price would be around 350 dollars. A Ti model is also in preparation and its technical sheet has caused a lot of reaction from players.

Laptops are no exception. It is rumored that the GeForce RTX 40XX for laptops will be launched in early 2023. The HP Omen 17 range has already leaked on a merchant site. The latter will offer versions of the GeForce RTX 4060, 4070, 4080 and 4090 with 13th generation Intel Core i7 processors. If the hallway rumors of social media are true, these big laptops will cost between 2000 € and 4000 €. Acer is also readying a high-end Acer Predator PH16-71 gaming laptop with a 24-core Core i9-13900HX Raptor Lake processor and a new NVIDIA RTX 4080 mobile graphics card. The latter has already been spotted on the Geekbench benchmark, and promises 40% more performance compared to the RTX 3080 mobile.

How much do the Nvidia GeForce RTX 4000s cost?

The GeForce RTX 4090 is Nvidia’s most powerful graphics card for consumers, and also the most expensive. In the French online catalog of Nvidia, it is now displayed at the price of 1849 euros minimum, for the Founder’s Edition (which is already out of stock on the official website). Custom versions from manufacturers Zotac and Gigabyte are also offered at this price, but also without stock. The other custom RTX 4090s are more expensive: count 2099.99 euros for a Zotac GAMING GeForce RTX 4090 Trinity, 2159.99 euros for a Gigabyte GeForce RTX 4090 GAMING OC 24G, 2189.99 euros for a Zotac GAMING GeForce RTX 4090 AMP Extreme AIRO and 2299.99 euros for an ASUS TUF-RTX4090-O24G-GAMING. But again, availability varies.

The price of the RTX 4080 starts at 1399 euros. But it may be difficult to find the graphics card at this price. It will probably be necessary to turn to more expensive custom models to succeed in acquiring them. The scalpers of course did not hesitate to monopolize the stocks to then resell them at exorbitant prices. For their part, Asus, MSI, Zotac and Gigabyte have already formalized RTX 4080s marketed under their brand. Only, for now, the graphics card is not really a success. Players shun it in favor of its big sister on shopping platforms.

The RTX 4070 Ti is expected to be the next generation graphics card to hit the market. Nvidia has announced a drop in its price for the United States: it will be offered there at $799 instead of $899. Will gamers from the Old Continent be entitled to such a discount? We can only hope so, but we cannot swear to anything. Currently, the price of the canceled RTX 4080 12GB replacement is estimated to start at 1099 euros.

The Nvidia GeForce RTX 4090 technical sheet

The GeForce RTX 4090 has a GPU made up of 16,384 CUDA cores with a base clock frequency of 2.23 GHz, for 2.52 GHz in boost. It also counts on the presence of 24 GB GDDR6X of memory and a 384-bit interface. It is of course PCI Express Gen 4 compatible.

The graphics card supports rendering in 4K HDR at 120Hz, 8K HDR at 60Hz, and variable refresh rates via its HDMI 2.1a connector. Its three DisplayPort 1.4a inputs enable display up to 4K 12-bit HDR at 240Hz with DP 1.4a + DSC, up to 8K 12-bit HDR at 60Hz with DP 1.4a + DSC or HDMI 2.1 + DSC and up to to 8K HDR at 120Hz with dual DP 1.4a + DSC.

The RTX 4090 consumes up to 450W and requires at least 850W of power to operate safely and at full power. Of course, some of this data may vary by graphics card model, with the information quoted taking the Founder’s Edition as an example. Note that in Undervolt, the RTX 4090 only loses 8% performance for an energy saving of 33%.

Nvidia’s most impressive graphics card is an overclocker’s dream. On November 21, 2022, for example, Team OGS, a team of graphics performance enthusiasts, managed to push the frequency of the GeForce RTX 4090 to 3.825 GHz with a special edition of the card designed by the Hong Kong brand Galax. It is obviously a a world recordwhich should soon be beaten as Nvidia’s GPU, in expert hands, lends itself well to this race for speed.

As far as performance is concerned, Nvidia promises a real leap forward with this generation thanks to the combination of even more advanced hardware and the integration of new technologies inherent in the Ada Lovelace architecture. According to Nvidia’s communication, an RTX 4090 performs almost twice as well as an RTX 3090 Ti in 4K quality and with DLSS Performance mode in Super Resolution on Flight Simulator, and more than twice as well on Warhammer 40,000: Darktide. The result is even more impressive on Cyberpunk 2077 in maximum ray tracing mode, which would run four times better on an RTX 4090 than on an RTX 3090 Ti thanks to the improvements in ray tracing.

The controversy over overheating problems

Be careful however, the start of the RTX 4090’s career is marred by overheating problems reported by several users. According to these multiple testimonials, some RTX 4090s heat up so much that they melt the power cable. While it was thought that this concern affected only a few unlucky owners, cases have multiplied in recent weeks. And what had to happen happened, a gamer decided to file a complaint against Nvidia after his graphics card overheated.

Nvidia GeForce RTX 4080 spec sheet

The GeForce RTX 4080 has 9728 Nvidia CUDA cores, for a clock frequency of 2.21 GHz that can go up to 2.51 GHz in boost. We also benefit from 16 GB of GDDR6X memory and a 256-bit bus size. PCI Express Gen 4 is included and the display capabilities of the connectors are the same as those of the RTX 4090.

The RTX 4080 can consume up to 320W and requires a computer with at least a 750W power supply to avoid unpleasant surprises.

Still according to Nvidia measurements and for a 4K definition, in DLSS Performance mode in Super Resolution, an RTX 4080 would enjoy almost twice the performance of an RTX 3080 Ti on Flight Simulator and Warhammer 40,000: Darktide. For Cyberpunk 2077 in maximum ray tracing mode, the RTX 4080 delivers more than three times the performance of the already excellent RTX 3080 Ti.

The already known Nvidia GeForce RTX 4070 Ti?

Nvidia has not yet presented its RTX 4070 Ti and has never even mentioned it yet, but the graphics card should resume the same technical characteristics as the late RTX 4080 12 GB, whose release has been canceled, but which should resume of the service under this name of RTX 4070 Ti.

The graphics card has 7680 CUDA cores, a clock frequency of 2.31 GHz (up to 2.61 GHz in boost), 12 GB of GDDR6X RAM, and a 192-bit memory bus. It develops 639 tensor-TFLOPs, 92 ray tracing-TFLOPs and 40 shader-TFLOPS. It requires a power supply of at least 700 W.

It is of course possible that Nvidia revises the data sheet of the product before its release.

GPU architecture Ada Lovelace: what’s new?

After the Turing architecture based on 12 nm engraving using TSMC’s FinFET process, introduced with the first RTX graphics cards (20 series), the Ampere architecture took over for the RTX 30 GPUs, based on a finesse of 8 nm etching developed by Samsung. The infidelity to TSMC did not last very long since Nvidia resumed business with the Taiwanese supplier for its Ada Lovelace architecture, which adopted a 4nm TSMC etching. This would make it possible to benefit from twice the energy efficiency of Ampere, that is to say that the GPUs can, for example, offer twice the performance for the same consumption, or offer the same performance when needed. half the energy.

Along with the Ada Lovelace architecture, also comes the DLSS 3.0, an evolution of Nvidia’s renowned upscaling technology. This new version of DLSS, which allows us to remember to significantly increase in-game performance by calculating a lower native definition, then performing an upgrade to the target definition by artificial intelligence, is for the moment exclusive to the RTX 40. The DLSS 3 is indeed powered by the new fourth-generation Tensor cores and the optical flow accelerator of the Ada Lovelace architecture. By predicting the geometry and movement of a scene, DLSS 3 reconstructs seven-eighths of the total pixels displayed, resulting in a substantial gain in in-game frames per second.

Ada Lovelace also integrates the function of Shader Execution Reordering (SER). This technology makes better use of GPU resources by reorganizing shading workloads, improving efficiency and potentially tripling ray tracing performance, as well as increasing in-game framerate by up to 25%.