NVIDIA has officially launched its new RTX 50 Series graphics cards, highlighted by the flagship model featuring 32 GB of GDDR7 memory and an impressive bandwidth of 1792 GB/s, delivering twice the performance of its predecessor. Priced at $1,999, this new model will be available starting January 30, 2025.
The formal announcement took place during a recent event in the USA, where NVIDIA presented a total of four models in the RTX 50 Series: the flagship model, along with the RTX 5080, RTX 5070 Ti, and RTX 5070. This new series is distinguished by the introduction of the Blackwell architecture, which significantly enhances performance, memory capacity, and processing power.
The flagship model is powered by the Blackwell GB202-300-A1 GPU, featuring 21,760 CUDA cores, marking a new milestone for the GeForce series by surpassing the 20,000-core threshold for the first time. Although this GPU utilizes 170 Streaming Multiprocessors (SM), which is an 11.4% decrease compared to the full configuration, its optimized architecture still allows it to outperform previous models significantly. The card’s robust memory operates on a 512-bit interface with a speed of 28 Gbps, achieving exceptional bandwidth.
Designed with a compact two-slot format, the flagship model incorporates a new 3D vapor chamber cooling system and a streamlined 14-layer PCB. Despite a high power requirement of up to 575W, the graphics card retains a compact size suitable for standard desktop setups.
In addition to the flagship, NVIDIA introduced the RTX 5080, which serves as the second high-end offering in the series. This model features the Blackwell GB203-400 GPU with 10,752 CUDA cores, 16 GB of GDDR7 memory, and a 256-bit interface. Notably, it is the only model in the RTX 50 Series that supports memory speeds of 30 Gbps, equating to a bandwidth of 960 GB/s. The power consumption for this model is rated at 360W, slightly higher than that of the RTX 4080, and it is priced at $999.
The RTX 5070 targets general users as the entry-level option in the RTX 50 Series. It is equipped with the GB205-300 GPU, which consists of 6,144 CUDA cores and 12 GB of GDDR7 memory, utilizing a 192-bit interface. This model achieves a bandwidth of 672 GB/s and has a power consumption of 250W, with pricing details to be confirmed upon release alongside the RTX 5070 Ti.
NVIDIA has placed a strong emphasis on enhancing heat dissipation and optimizing design for its Founders Edition models. The innovative PCB design incorporates back drill technology along with a Vapor Chamber 3D heatsink system, greatly improving cooling efficiency while maintaining a compact form factor. This design focus is particularly crucial for the flagship model, which demands effective cooling solutions to sustain high performance over extended periods.
The RTX 50 Series is set to make its official debut in early 2025, with the flagship model and RTX 5080 launching on January 30, followed by the RTX 5070 Ti and RTX 5070 in February. With remarkable performance, cutting-edge technology, and a variety of pricing options, this generation of graphics cards represents a significant advancement for NVIDIA in the global GPU market.