Those attending CES 2025 have been exceptional witnesses of the presentation of the new generation of GeForce RTX 50 graphics cardsbased on NVIDIA’s Blackwell architecture. Luckily, you don’t have to visit the largest technology fair in the world to see how the entire event unfolded and discover all the details about the announced models and their respective prices.
Extreme performance with the RTX 5090
At the head of the line stands a powerful GeForce RTX 5090which comes equipped with 32 GB of VRAM memory in the new GDDR7 protocol. It promises a significant jump in performance, indicative of high energy consumption. The RTX 5090 impresses with its 4,000 TOPS AI (well above the 1,300 TOPS of the RTX 4090) and 1.8 TB/s bandwidth, enabling higher frame rendering rates in games and professional tasks. These technologies reinforce the brand’s commitment to driving much higher performance, especially in applications that require artificial intelligence.
Introducing GeForce RTX 50 Series graphics cards & laptops. Powered by NVIDIA Blackwell architecture & AI, the RTX 50 Series enables new experiences & next-level graphics fidelity.
Learn More → https://t.co/2wsweKcWTapic.twitter.com/bRx1Z6qEpR
— NVIDIA GeForce (@NVIDIAGeForce) January 7, 2025
RTX 5090 – Features:
CUDA Cores: 21,760
Tensor Cores (AI): 3,352 AI TOPS
Ray Tracing Cores: 318 TFLOPS
Base Clock: 2.01 GHz
Boost Clock: 2.41 GHz
Memory: 32GB GDDR7
Memory interface: 512-bit
TGP: 575W
Additionally, the GeForce RTX 5090 features an angled 12 VHPWR connector on the side, suggesting high power demands. Even so, the company has not revealed complete data on the consumption of the most robust card in the family. It will sell for $1,999.
GeForce RTX 5080 and RTX 5070
We now move on to the GeForce RTX 5080, an option aimed at the mid-range segment which comes equipped with 16 GB of GDDR7 memory, offering a powerful solution for gamers and content creators looking for high performance. It will sell for $999.
RTX 5080 – Features:
CUDA Cores: 10,752
Tensor Cores (AI): 1,801 AI TOPS
Ray Tracing Cores: 171 TFLOPS
Base Clock: 2.30 GHz
Boost Clock: 2.62 GHz
Memory: 16GB GDDR7
Memory interface: 256-bit
TGP: 360W
On the other hand, the models were also presented GeForce RTX 5070 Ti and GeForce RTX 5070which expand their catalog to serve both casual players and professionals looking for high performance. The first will initially be sold for $749, while its basic version will be reduced to $549.
RTX 5070 Ti – Features:
CUDA Cores: 8,960
Tensor Cores (AI): 1,406 AI TOPS
Ray Tracing Cores: 133 TFLOPS
Base Clock: 2.30 GHz
Boost Clock: 2.45GHz
Memory: 16GB GDDR7
Memory interface: 256-bit
TGP: 300W
RTX 5070 – Features:
CUDA Cores: 6,144
Tensor Cores (AI): 988 AI TOPS
Ray Tracing Cores: 94 TFLOPS
Base Clock: 2.16 GHz
Boost Clock: 2.51 GHz
Memory: 12GB GDDR7
Memory interface: 192-bit
TGP: 250W
It is worth noting that these prices stipulated in dollars correspond to North American territory and that the new models of graphics cards Their price may vary in Europe due to import taxes and other costs. In any case, his arrival is scheduled for first quarter of 2025 with the premiere of the 5090 and 5080 models on January 30, while the two versions of the 5070 will arrive in February.
GeForce RTX 50 architecture
We are now focusing on the introduction of GDDR7 memory, which guarantees faster transfer speeds, significantly improving performance in graphics-intensive games and applications. The Blackwell architecture has been developed to optimize energy consumption, guaranteeing an efficient balance between power and savings.
NVIDIA is also introducing RTX Neural Shaders technology at the launch of GeForce RTX 50 GPUs, integrating AI networks into programmable shaders. This feature promises to raise the quality of materials, lighting and other graphic effects in real time, reaching a standard close to cinematic. In this area, RTX Neural Faces stands out, a tool designed to optimize the representation of characters in games. Using generative AI, this feature transforms simple rasterized face and 3D pose data into high-quality digital representations that are stable over time.
New technologies
Another element to highlight in the GeForce RTX 50 series is DLSS 4which will increase gaming performance up to 8 times, generating up to three additional frames for every frame rendered using artificial intelligence. This is accompanied by Reflex 2, which introduces an advanced technique to reduce latency, allowing the frame to be updated based on the last click of the controller, mouse or keyboard before being monitored.
With these advances, the technology house reaffirms its position in GPU technology, setting a new standard in the high-end graphics card market and consolidating the GeForce RTX 50 as a reference for gamers.