
NVIDIA GeForce RTX 5070 Ti & RTX 5070 Specifications Leak | Image Source: wccftech.com
SANTA ​CLARA, California, December 25, 2024 – NVIDIA’s Blackwell RTX 50 series continues to make waves before its ​official unveiling at CES 2025, with new leaks detailing the GeForce RTX 5070 Ti and RTX 5070 ​Ti specifications. ​According to the famous Kopite7kimi Filter, the new GPU is designed to redefine the performance parameters of high-end cards.
According to Wccftech, the RTX 5070 Ti will have the death of GB203, with 8,960 Cuda nuclei and 16GB ​of GDDR7 memory on a 256-bit bus. Meanwhile, the RTX 5070 should use the death of GB205, with 6,144 Cuda nuclei ​and 12 GB GDDR7 memory on a ​192-bit bus. Both GPUs are adapted to offer exceptional gaming ​and productivity experiences while maintaining efficiency.
Specifications of the RTX ​5070 Ti
The RTX 5070 Ti stands out as a powerful competitor in the next line of NVIDIA. It is built on the GB203-300-A1 matrix and ​incorporates 8,960 ​CUDA nuclei. According to Kopite7kimi, this GPU offers 16 GB of GDDR7 memory on a 256-bit memory interface. With a ​total mounting power of 300W, it is designed to provide robust performance and improve the energy efficiency of its predecessors.
The industry’s interior highlights the 50W ​reduction ​in TBP compared to previous speculations. ​This brings RTX 5070 Ti energy consumption closer to RTX 4070 Ti, which allows to ​balance high performance with lower energy demand. As Wccftech pointed out, although most models probably adhere to TBP 300W, some ​variants could choose a higher ​350W configuration.
“The RTX 5070 The combination of Cuda hearts, ​VRAM and energy efficiency makes it a leading GPU for players and professionals,” said Kopite7kimi.
Specifications of RTX 5070
The GeForce RTX 5070 serves as ​a workhorse of the series, offering a competitive balance of power and ​affordability. It uses the GB205-300-A1 matrix, optimized for ​6144 CUDA nuclei. The GDDR7 12GB memory configuration works ​on a ​192-bit bus, which coincides ​with its predecessor, the RTX 4070, but presents GDDR7 memory to improve bandwidth. This GPU has a TBP 250W, ensuring efficiency without ​compromising performance.
According to Kopite7kimi, the number of ​Cuda nuclei marks a slight decrease in earlier rumours ​of ​6,400 nuclei. However, the inclusion of the GDR7 memory should increase the overall performance of the data, which will improve the performance ​of the games and increase the workload.
Context of the RTX 50 series
The GeForce RTX 5070 Ti and the ​RTX 5070 are positioned as part of NVIDIA’s Blackwell architecture, an important step in the RTX 50 line. Other models, including ​the RTX 5090 and the RTX 5080, are expected to start in the first quarter of 2025. These GPUs promise considerable performance gains, with the RTX ​5090 providing 21,760 ​CUDA canes and a stunning ​32GB GDDR7 memory on a 512-bit bus.
The ​prices ​of RTX ​5070 Ti and RTX 5070 remain secret, but analysts predict that the costs will be closely aligned with their RTX 40 series counterparts. As Wccftech has indicated, it is unlikely that NVIDIA will reduce prices, with the aim of offering value through performance improvements and technological advances such as the memory of the ​GDR7 and ​the improvement of Cuda’s hearts.
Comparison with precedents
The RTX 5070 configuration Ti 16 GB VRAM represents a significant improvement over the RTX 4070 Ti, which had 12 GB of GDDR6X memory. Similarly, the RTX 5070 introduces GDDR7 memory but retains the RTX 4070’s ​12GB capacity. These updates promise to increase the ​game and creative ​workflows, especially in managing 4K resolutions and AI-based applications.
Efficiency gains are also remarkable. While the RTX 4070 Ti and RTX 4070 had 285W and 200W TBB ​respectively, their successors maintain higher performance with relatively modest increases in energy consumption, as highlighted by the filtered specifications.
Date of launch and expectations
Both ​RTXs should start 5070 Ti and RTX 5070 shortly after ​the presentation of the RTX 5080 and ​RTX 5090s in ​early 2025. ​According to Kopite7kimi, ​the initial launch will prioritise high-end models, with GPUs of medium range and entry level ​depending on the combination. This strategy is consistent with the traditional NVIDIA broadcast cycles, ensuring ​a broad portfolio that meets the different needs of users.
The CES 2025 event should be a crucial moment for NVIDIA, the RTX 50 series taking the central stage. Industry experts anticipate that these GPUs will set new standards for performance, efficiency and ​innovation, NVIDIA’s continuing field in the graphic card market.
While leaks offer an exciting look at NVIDIA plans, official specifications and prices will finally determine the success of the ​RTX 5070 Ti ​and RTX 5070. With the RTX 50 series to redefine professional games and graphics, ​all eyes are in NVIDIA as the new year approaches.