Nvidia expect a 2-4x performance boost with their RTX 40 series
Published: 20th September 2022 | Source: Nvidia | Author: Mark Campbell
Nvidia blasts though expectations with their Ada RTX 40 series GPU launch
Nvidia has just revealed their RTX 4090 and RTX 4080, the company's first graphics cards to utilise their new Ada Lovelace GPU architecture and TSMC's 4N manufacturing process.
With their RTX 40 series of graphics cards Nvidia has boasted 2x-4x performance boosts over their RTX 30 series delivering to gamers huge performance leaps and huge performance-per-watt benefits. Yes, Nvidia's RTX 4090 has a 450W TDP, but Ada can deliver enough performance to make those high wattages seen reasonable.
With new tensor cores, RT cores, and shader engines, Nvidia's RTX 40 series can deliver gamers a transformative leap in computational performance, be it for traditional rendering, ray tracing, or AI workloads. Nvidia's Ada architecture is a huge leap over Ampere, and that's a big deal for gamers.
RTX 4090
Nvidia's RTX 4090 graphics card will become available on October 12th with 24GB of GDDR6X memory. This graphics card is said to be 2-4 times faster than Nvidia's RTX 3090 Ti, a performance feat that makes the graphics card's $1.599 price tag seem a lot less egregious than its last-generation counterpart.
Nvidia's RTX 4090 graphics card features over 76 billion transistors, 3rd generation RT cores, 4th generation Tensor cores, and supports the company's upcoming DLSS 3.0 technology. This GPU offers users single-precision performance of 82.6 TFLOPS, which is 2.3x more than its predecessor RTX 3090, and features over 1TB/s of memory bandwidth.
Nvidia's RTX 4090 will feature 16,384 CUDA cores, a boost clock speed of 2.52 GHz, and support Resizable BAR out of the box. The GPU is 3-slots, 61mm, thick, 304mm long, and 137mm wide.
RTX 4080
With Nvidia's RTX 4080, Nvidia promises a 2-4X performance boost over Nvidia's RTX 3080 Ti. The graphics card will become available in November in two flavours, a $899 12GB version, and a $1,199 16GB option. Both models will feature GDDR6X memory.
With a 2-4x performance boost over Nvidia's RTX 3080 Ti, both of these graphics cards will be more powerful than Nvidia's RTX 3090 Ti graphics cards. This GPU will use a AD103 graphics chip.
Both Nvidia's RTX 4090 and RTX 4080 GPUs will utilise PCIe 4.0 connectivity, which means that Nvidia's RTX 40 series will lack support for PCIe 5.0. Nvidia's RTX 40 series GPUs will support Nvidia's 8th generation NVENC encoder and support AV1 encode and decode operations.
RTX 4080 16GB VS RTX 4080 12GB
Nvidia's RTX 4080 16GB doesn't just contain more VRAM than its 12GB cousin, it also features more than 20% more CUDA cores. The RTX 4080 16GB will feature 9728 CUDA cores, and the RTX 4080 12GB has only 7680. It could be argued that Nvidia's RTX 4080 16GB should be called the RTX 4080 Ti, or that that RTX 4080 12GB should be called the RTX 4070.
Below are the official specifications of Nvidia's RTX 4090, RTX 4080 16GB, and RTX 4080 12GB
DLSS 3.0
Nvidia has also today revealed DLSS 3.0, promising uses higher framerates in both CPU-limited games and GPU-limited games. Below is how DLSS 3.0 can impact the performance of Microsoft Flight Simulator, one of the most CPU-limited games on the market.
With DLSS 3.0, Nvidia uses their fourth-gen Tensor Cores and Optical Flow Accelerator on GeForce RTX 40 Series GPUs to generate intermediate frames to effectively double the performance of supported games.
You can join the discussion on Nvidia's RTX 4090 and RTX 4080 graphics cards on the OC3D Forums .