NVIDIA 40 Series Graphics Cards: What we know so far! 

NVIDIA RTX 40 Series

NVIDIA CEO Jensen Huang finally announced the news we’ve all been waiting for. The new 40 Series GPUs are launching this autumn! Three cards were announced during the GeForce Beyond Special Broadcast event: the GeForce RTX 4090, and two versions of the GeForce RTX 4080.

Ada Lovelace, the Ultra Efficient Architecture

To power this new generation of GPUs, NVIDIA has unveiled its 3^rd generation of RTX architecture, Ada Lovelace. So named after Augusta Ada King, Countess of Lovelace, who is often regarded as the first programmer. With this new architecture, the 40 Series GPUs will give gamers and content creators a huge leap in performance, with more immersive gaming experiences and super-fast content creation workflow.

The NVIDIA Ada Lovelace architecture is built upon a custom TSMC 4N process, that delivers 76 billion transistors compared to the 28 billion of the previous generation. This massive leap in capabilities allows Ada to be incredibly efficient, delivering twice the performance compared to Ampere. Furthermore, Ada provides excellent scalability and overclocking ability per watt delivered when compared to previous generations.

Alongside this incredible new process, NVIDIA’s Ada also delivers a boost to your games thanks to an increase in Streaming Multiprocessors (SMs). NVIDIA previously employed double peak FP32 throughput with the last generation. They are now more than doubling the throughput with Ada. NVIDIA is packing up to 90-TFLOPS of Shader power on the 4000 Series, with the GeForce RTX 4090 boasting an impressive 83-TFLOPS.

Next-Gen Cores

With the 40 series of GPUs comes the latest generation of Ray Tracing and Tensor cores. NVIDIA have utilised their third generation of Ray Tracing cores, which build upon the incredible performance of previous iterations. These next-gen RT cores have twice the ray-triangle intersection throughput, offering an increase in RT-TFLOP performance of up to 2.8 times. The GeForce RTX 4090 will offer 191 RT-TFLOPS, compared to the 78 RT-TFLOPs of the fastest GPU of the previous generation, enabling immersive ray traced worlds to be rendered much faster.

The prior generation of NVIDIA GPUs was also known for its AI capabilities, which powered the impressive Deep Learning Super Sampling (DLSS). To leverage this capability even further, NVIDIA has incorporated their 4^th generation of Tensor cores into Ada. These cores are unbelievably fast, with a new 8-bit Floating Point (FP8) Tensor engine. This offers up to five times increased throughput, with the GeForce RTX 4090 offering 1.32 petaFLOPS!

Innovative Technologies

To harness these advanced cores, NVIDIA has included several innovations in the 4000 Series. To more efficiently organise the workload of ray tracing, they have included a new technology called Shader Execution Reordering (SER). This offers up to twice the shader performance, and an increase of in-game frame rates up to 25%!

Alongside their excellent gaming capabilities, these cards will be great for live streamers and content creators. This is thanks to utilising two of their eighth-generation NVIDIA Encoders (NVENC) which support AV1 encoding. Live streamers can enjoy improved encoding efficiency by 40%, meaning live streams will look like they’re using 40% more bitrate than they do currently. Furthermore, the dual encoders are up to twice as fast, ensuring creators will be able to save time on their projects.

Finally, NVIDIA has included DLSS 3 with this new architecture. Integrated within DLSS 3 is a new AI-powered Optical Multi Frame Generation technology. Through a process that combines DLSS Super Resolution, DLSS Frame Generation, and NVIDIA Reflex, DLSS 3 is able to reconstruct 7.8ths of the displayed pixels for entirely new frames, increasing performance significantly. In games that rely heavily upon the GPU, DLSS 3 can boost framerates by up to four times, and even CPU-limited games can see frame rates improved by up to two times.

Meet the 40 Series Cards

The top-of-the-line GPU announced during the GeForce Beyond Special Broadcast is the GeForce RTX 4090, due to launch on October 12th. With this graphics card in your system, you’ll have access to 16,385 CUDA Cores, 1,321 Tensor T-FLOPS, 191 RT-TFLOPS, and 83 Shader T-FLOPS of power, with 24GB of G6X VRAM. With so much performance, you’ll easily be able to play the latest games at the highest settings with spectacular levels of immersion. In today’s ray-traced games you can see twice the performance, and by leveraging innovations such as DLSS 3 and SER you can expect even up to four times the performance.

Alongside the impressive GeForce RTX 4090, NVIDIA also announced two versions of the GeForce RTX 4080, due for release in November. The GeForce RTX 4080 (16GB) boasts 9,728 CUDA Cores, 780 Tensor T-FLOPs, 113 RT-TFLOPs, 49 Shader-TFLOPs, and 16GB of GDDR6X VRAM.  Meanwhile the GeForce RTX 4080 (12GB) has 7,680 CUDA Cores, 639 Tensor-TFLOPs, 92 RT-TFLOPs, 40 Shader-TFLOPs, and 12 GB of GDDR6X VRAM. Making these cards capable of outperforming the powerful GeForce RTX 3090 Ti!

GeForce RTX 40 Series Specifications

In this table you will find the technical breakdowns for each of the announced 40 Series GPUs, and how they compare to the elite-tier 30 Series GPUs. 

New Levels of Detail

We here at Overclockers UK can’t wait to see what will be possible with this new generation of GPUs. NVIDIA has already been demonstrating what they are capable of, such as with games that utilise RTX technology, for example Cyberpunk 2077. They also announced an exciting DLC RTX update to the classic Portal. Many upcoming games such as Overwatch 2, Warhammer 40,000: Darktide, and Call of Duty: Modern Warfare II are also sure to impress us with stunning visuals.