Las Vegas – January 6, 2025 – After many months of speculations and rumours leading up to its release, the next generation of GeForce RTX consumer desktop and laptop GPUs have been officially announced. These include the RTX 5090, RTX 5080, RTX 5070 Ti and the RTX 5070, powered by the incredible new Blackwell architecture featuring 5th Gen Tensor Cores and 4th Gen Ray Tracing Cores.
Taking to the stage at CES 2025, Jensen Huang (founder and CEO of NVIDIA) wasted no time in unveiling the RTX 50 series lineup before delving into many other exciting forays that NVIDIA has been making in other areas as well.
“Blackwell, the engine of AI, has arrived for PC gamers, developers and creatives. Fusing AI-driven neural rendering and ray tracing, Blackwell is the most significant computer graphics innovation since we introduced programmable shading 25 years ago.”
Jensen Huang, founder and CEO of NVIDIA
In line with many speculations, the RTX 5090 features an impressive 32 GB of VRAM coupled with 680 Tensor Cores, 170 RT Cores and a TBP of 575, utilising the GB202 architecture on a 4 nm process node. In total it comprises of a whole 92 billion transistors capable of providing over 3,352 trillion AI operations per second (TOPS) of computing power and comes with the newly announced DLSS 4, supposedly further boosting its performance by up to 2x the Ada Lovelace powerhouse that is the RTX 4090.
We’ve compiled a list of all the available data for these for all the announced cards below:
| Specs | RTX 5090 | RTX 4090 | RTX 5080 | RTX 4080 S | RTX 5070 Ti | RTX 4070 Ti S | RTX 5070 | RTX 4070 S |
|---|---|---|---|---|---|---|---|---|
| Architecture | GB202 | AD102 | GB203 | AD103 | GB203 | AD103 | GB205 | AD104 |
| Process Node | N4P | N5 | N4P | N5 | N4P | N5 | N4P | N5 |
| SMs | 170 | 128 | 84 | 80 | 70 | 66 | 48 | 56 |
| CUDA Cores | 21760 | 16384 | 10752 | 10240 | 8960 | 8448 | 6144 | 7168 |
| Tensor Cores | 680 | 512 | 336 | 320 | 280 | 264 | 192 | 224 |
| Ray Tracing Cores | 170 | 128 | 84 | 80 | 70 | 66 | 48 | 56 |
| Base Clock (GHz) | 2.01 | 2.23 | 2.3 | 2.29 | 2.3 | 2.34 | 2.16 | 1.98 |
| Boost Clock (GHz) | 2.4 | 2.5 | 2.62 | 2.55 | 2.45 | 2.61 | 2.51 | 2.47 |
| VRAM (GB) | 32 G7 | 24 G6X | 16 G7 | 16 G6X | 16 G7 | 12 G6X | 12 | 12 |
| Bus Width (bit) | 512 | 384 | 256 | 256 | 256 | 256 | 192 | 192 |
| Bandwidth (GB/s) | 1792 | 1008 | 960 | 736 | 896 | 672 | 672 | 504 |
| Texture Units | 680 | 512 | 336 | 320 | 280 | 264 | 192 | 224 |
| TFLOPS FP32 | 104.8 | 82.58 | 56.3 | 52.22 | 43.9 | 44.10 | 30.9 | 35.48 |
| TBP | 575W | 450W | 360W | 320W | 300W | 285W | 250W | 220W |
| Rec PSU | 1000W | 850W | 850W | 700W | 750W | 600W | 650W | 550W |
The RTX 5090 and the RTX 5080 are scheduled to be available starting January 30th, while the RTX 5070 Ti and RTX 5070 will be available starting February. Have a look at the announced prices for India:
| 50 Series GPU | RTX 5090 | RTX 5080 | RTX 5070 Ti | RTX 5070 |
|---|---|---|---|---|
| Pricing | INR 2,14,000 | INR 1,07,000 | INR 80,000 | INR 59,000 |
| Availability | January 30th 2025 | January 30th 2025 | February 2025 | February 2025 |
The prices are surprisingly lesser than the initial rumours floating around, with the RTX 5090 reaching INR 3,00,000 in some estimates, and with the RTX 5070 claimed to be able to reach the performance of the RTX 4090, granted with AI boosts and not raw rasterisation, it is still a pretty impressive lineup overall.
The RTX 50 series also comes with a host of AI powered improvements to further enhance its performance by a significant amount. We’ll dive deeper on a separate post but here’s a quick overview:
- DLSS 4 which debuts Multi Frame Generation to boost frames by using AI to generate up to three frames per rendered frame, meaning every 4th frame rasterised.
- NVIDIA Reflex 2, which introduces Frame Warp. This technique reduced latency in games by updating a rendered frame based on the latest mouse input just before it is sent to the display.
- RTX Neural Shaders, which brings small AI networks into programmable shaders
- RTX Neural Faces which using generative AI to render temporally stable, high quality digital faces in real time.
With this reveal, NVIDIA has solidified its stance on using cutting edge AI features to boost gaming performance and bring accessibility to gaming at affordable prices, which would have been considered impossible otherwise in the age of pure rasterisation. There is a debate to be made between pure rasterised frames and AI powered ones, but in the end it is just another tool to enhanced the price-performance ratio and if it works, it works. And so far DLSS has been a pretty impressive tool.
Stay tuned for more CES updates and deeper dives into some of the other announced technologies from NVIDIA and other competitors.
