Hot on the heels of VRAM specs leaks, further GeForce RTX 50 Series rumours have arrived. These whispers concern GeForce RTX 5070 Ti, namely its TBP and GPU. Like its predecessor, GeForce RTX 4070 Ti Super, it’ll apparently be closely related to the next-most-powerful card in the family, GeForce RTX 5080, but will require more power to realise its performance.
According to sources close to Wccftech, both GeForce RTX 5070 Ti and GeForce RTX 5080 will each use a GB203 GPU. More specifically, the latter will carry the full-fat GB203-400 with 10,752 CUDA cores whereas the former will have a cutdown GB203-300 at its heart and 8,960 CUDA cores at its disposal. This accounts for a potential 17% shortfall between the two cards, and a minor 6% bump relative to GeForce RTX 4070 Ti Super. Of course, this doesn’t account for any architectural differences.
Since rumours claim GeForce RTX 5070 Ti sports the same GPU as GeForce RTX 5080, memory bus width should remain the same at 256-bit. In other words, this card should carry 16GB of 28Gbps GDDR7 VRAM, likely totalling 896GB/s of bandwidth. If correct, this would result in a 33% uplift over GeForce RTX 4070 Ti Super.
Lastly, these same sources also claim that GeForce RTX 5070 Ti TBP is 350W, up 65W from GeForce RTX 4070 Ti Super. If true, it’s unfortunate to see Nvidia taking a step back on power consumption, even if the improvement last generation was relatively minor at 5W. That said, leaker @kopite7kimi has indicated that while 350W is one of the possible configurations, latest data points to a more reasonable 285W target. Here’s hoping for the latter.
All of this should become clear soon as Nvidia is announces it GeForce RTX 50 Series on January 7 at 2:30AM (GMT). So, until then take the above with a grain of salt.
