Nvidia unveils GeForce RTX 40 collection graphics playing cards with 4X efficiency

Nvidia CEO Jensen Huang took the wraps off the corporate’s newest graphics playing cards immediately — the consumer-focused GeForce RTX 40 Sequence of graphics processing items (GPUs).

The collection for avid gamers and creators options the flagship chip, the RTX 4090 GPU, with as much as 4 instances the efficiency of its predecessor.

These chips and graphics playing cards are the primary GPUs primarily based on the brand new Nvidia Ada Lovelace structure. Huang promised huge generational leaps in efficiency and effectivity, saying it represents a brand new period of real-time ray tracing and neural rendering, which makes use of AI to generate pixels.

“The age of RTX ray tracing and neural rendering is in full steam, and our new Ada Lovelace architecture takes it to the subsequent degree,” stated Huang through the GeForce Past: Particular Broadcast on the firm’s fall GTC occasion. “Ada supplies a quantum leap for avid gamers and paves the way in which for creators of absolutely simulated worlds. With as much as 4x the efficiency of the earlier technology, Ada is setting a brand new commonplace for the trade.”

Huang confirmed off a video dubbed Racer X that confirmed toy automobiles racing round a room with superb shadows and lighting. Huang stated Ada Lovelace is a large leap over the earlier technology.

“Ada, proper off the bat, supplies about 2X efficiency enchancment on immediately’s video games,” stated Matt Wuebbling, international head of Nvidia GeForce advertising and marketing, in a press briefing. “That’s fairly unparalleled in an structure launch. That is our largest leap ever. After which there may be DLSS3, which is the AI part of it.”

He stated the Racer X video that confirmed off the simulation capabilities of the GPU with a few 4X enchancment.

“Racer X is a simulation,” Huang stated. “Racer X is working on one single GPU.”

DLSS 3 generates whole frames for quicker recreation play

Huang additionally introduced Nvidia DLSS 3 — the subsequent revolution within the firm’s Deep Studying Tremendous Sampling neural-graphics expertise for video games and artistic apps.

The AI-powered expertise can generate whole frames for massively quicker recreation play. It could possibly overcome CPU efficiency limitations in video games by permitting the GPU to generate whole frames independently. The AI can predict what sorts of pixels to fill in quicker in a shifting scene.

The expertise is coming to the world’s hottest recreation engines, equivalent to Unity and Unreal Engine, and has obtained help from lots of the world’s main recreation builders, with greater than 35 video games and apps coming quickly, Nvidia stated.

• Moreover, the RTX 40 Sequence GPUs characteristic a variety of latest technological improvements, together with: Streaming multiprocessors with as much as 83 teraflops of shader energy — 2x over the earlier technology.
• Third-generation RT Cores with as much as 191 efficient ray-tracing teraflops — 2.8x over the earlier technology.
• Fourth-generation Tensor Cores with as much as 1.32 Tensor petaflops — 5x over the earlier technology utilizing FP8 acceleration.
• Shader Execution Reordering (SER) that improves execution effectivity by rescheduling shading workloads on the fly to higher make the most of the GPU’s sources. As important an innovation as out-of-order execution was for CPUs, SER improves ray-tracing efficiency as much as thrice and in-game body charges by as much as 25%.
• Ada Optical Circulate Accelerator with 2x quicker efficiency permits DLSS 3 to foretell motion in a scene, enabling the neural community to spice up body charges whereas sustaining picture high quality.
• Architectural enhancements tightly coupled with customized TSMC 4N course of expertise leads to an as much as 2x leap in energy effectivity.
• Twin Nvidia Encoders (NVENC) lower export instances by as much as half and have AV1 help. The NVENC AV1 encode is being adopted by OBS, Blackmagic Design DaVinci Resolve, Discord and extra.

New ray-tracing for immersive video games

For many years, rendering ray-traced scenes with bodily appropriate lighting in actual time has been thought of the holy grail of graphics. On the identical time, geometric complexity of environments and objects has continued to extend as 3D video games and graphics try to offer essentially the most correct representations of the actual world, Nvidia stated.

Attaining bodily correct graphics requires super computational horsepower. Fashionable ray-traced video games like Cyberpunk 2077 run over 600 ray-tracing calculations for every pixel simply to find out lighting — a 16x improve from the primary ray-traced video games launched 4 years in the past.

The brand new third-generation RT Cores have been enhanced to ship 2x quicker ray-triangle intersection testing and embody two essential new {hardware} items. An Opacity Micromap Engine accelerates ray tracing of alpha-test geometry by an element of 2x, and a Micro-Mesh Engine generates micro-meshes on the fly to generate extra geometry. The Micro-Mesh Engine supplies the advantages of elevated geometric complexity with out the normal efficiency and storage prices of advanced geometries.

“Nvidia engineers pushed expertise on each entrance,” Huang stated.

RTX Remix and AV1 Encoders

The RTX 40 Sequence GPUs and DLSS 3 ship developments for Nvidia Studio creators. 3D artists can render absolutely ray-traced environments with correct physics and real looking supplies, they usually can view the adjustments in actual time, with out proxies.

Video modifying and stay streaming additionally get a lift from improved GPU efficiency and the inclusion of latest twin, eighth-generation AV1 encoders. The Nvidia Broadcast software program improvement package has three updates, now accessible for companions, together with Face Expression Estimation, Eye Contact and high quality enhancements to Digital Background.

Nvidia Omniverse — included within the Nvidia Studio suite of software program — will quickly add Nvidia RTX Remix, a modding platform to create beautiful RTX remasters of traditional video games. RTX Remix permits modders to simply seize recreation belongings, routinely improve supplies with highly effective AI instruments, and shortly allow RTX with ray tracing and DLSS.

RTX Remix has been utilized by Nvidia Lightspeed Studios to reimagine Valve’s online game Portal, considered among the best video video games of all time. Superior graphics options equivalent to full ray tracing and DLSS 3 give the sport a placing new appear and feel. Portal with RTX will likely be launched as free, official downloadable content material for the traditional platformer with RTX graphics in November, simply in time for Portal’s fifteenth anniversary.

“We created an Omniverse software known as RTX Remix,” Huang stated.

The GeForce RTX 4090 and 4080 GPUs

The RTX 4090 is the world’s quickest gaming GPU with astonishing energy, acoustics and temperature traits. In full ray-traced video games, the RTX 4090 with DLSS 3 is as much as 4x quicker in comparison with final technology’s RTX 3090 Ti with DLSS 2.

Additionally it is as much as two instances quicker in immediately’s video games whereas sustaining the identical 450W energy consumption. It options 76 billion transistors, 16,384 CUDA cores and 24GB of high-speed Micron GDDR6X reminiscence, and persistently delivers over 100 frames per second at 4K-resolution gaming. The RTX 4090 will likely be accessible on October 12, beginning at $1,600.

The corporate additionally introduced the RTX 4080, launching in two configurations. The RTX 4080 16GB has 9,728 CUDA cores and 16GB of high-speed Micron GDDR6X reminiscence, and with DLSS 3 is 2x as quick in immediately’s video games because the GeForce RTX 3080 Ti and extra highly effective than the GeForce RTX 3090 Ti at decrease energy.

The RTX 4080 12GB has 7,680 CUDA cores and 12GB of Micron GDDR6X reminiscence, and with DLSS 3 is quicker than the RTX 3090 Ti, the previous-generation flagship GPU. Each RTX 4080 configurations will likely be accessible in November, with costs beginning at $1,200 and $900, respectively.

The place to purchase

The GeForce RTX 4090 and 4080 GPUs will likely be accessible as customized boards, together with stock-clocked and factory-overclocked fashions, from high add-in card suppliers equivalent to Asus, Colourful, Gainward, Galaxy, GIGABYTE, Innovision 3D, MSI, Palit, PNY and Zotac.

The RTX 4090 and RTX 4080 (16GB) are additionally produced immediately by Nvidia in restricted Founders Editions for followers wanting the Nvidia in-house design.

The GeForce RTX 40 Sequence GPUs will likely be in gaming techniques constructed by Acer, Alienware, ASUS, Dell, HP, Lenovo and MSI, main system builders worldwide, and lots of extra. And keep in mind what Huang at all times says. The extra you purchase, the extra you save.