Extremely anticipated: Nvidia has formally unveiled its next-gen GeForce RTX 40 sequence GPUs based mostly on the brand new Ada Lovelace structure and headlined by the flagship RTX 4090. Ada Lovelace is constructed on TSMC’s 4N course of and can debut in new RTX 4090 and RTX 4080 graphics playing cards. The previous packs 76 billion transistors, 16,384 CUDA cores, and 24GB of GDDR6X reminiscence from Micron.

Based on Nvidia, an RTX 4090 with DLSS 3 is as much as 4 instances quicker than an RTX 3090 Ti working DLSS 2, and effectivity positive factors imply it sticks with the identical 450W energy consumption because the RTX 3090 Ti.

DLSS 3 is Nvidia’s newest Deep Studying Tremendous Sampling neural-graphics know-how. It’s constructed on DLSS Tremendous Decision, and provides Optical Multi Body Technology to create completely new frames to considerably enhance body charges.

Nvidia has a deeper dive accessible on its web site, however it is best to know that the tech is unique to RTX 40 sequence GPUs and is coming to a number of the hottest recreation engines together with Unreal Engine and Unity. Greater than 35 video games and apps are already within the pipeline with the primary coming in October.

Nvidia has introduced two flavors of the GeForce RTX 4080 GPU. The RTX 4080 12GB options 7,680 CUDA cores alongside 12GB of Micron GDDR6X reminiscence and with DLSS 3 enabled (vs DLSS 2), it is claimed to be quicker than the RTX 3090 Ti — it can set you again $899.

A better-spec model with 9,728 CUDA cores and 16GB of GDDR6X reminiscence may also be accessible, albeit for a bit extra coin at a hefty $1,199.

Some function and architectural highlights embody:

  • Streaming multiprocessors with as much as 83 teraflops of shader energy — 2x over the earlier era.
  • Third-generation RT Cores with as much as 191 efficient ray-tracing teraflops — 2.8x over the earlier era.
  • Fourth-generation Tensor Cores with as much as 1.32 Tensor petaflops — 5x over the earlier era utilizing FP8 acceleration.
  • Shader Execution Reordering (SER) that improves execution effectivity by rescheduling shading workloads on the fly to raised make the most of the GPU sources. As vital an innovation as out-of-order execution was for CPUs, SER improves ray-tracing efficiency as much as 3x and in-game body charges by as much as 25%.

  • Ada Optical Stream 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 know-how ends in an as much as 2x leap in energy effectivity.
  • Twin Nvidia Encoders (NVENC) minimize export instances by as much as half and have AV1 assist. The NVENC AV1 encode is being adopted by OBS, Blackmagic Design DaVinci Resolve, Discord and extra.

The flagship RTX 4090 launches on October 12 beginning at $1,599, with the RTX 4080 12GB and 16GB variants touchdown someday in November from $899 and $1,199, respectively. All will probably be accessible from add-in companions together with Asus, Gigabyte, MSI and Zotac, in each stock-clocked and manufacturing unit overclocked variants.