Nvidia Blackwell Rumors: GeForce RTX 5070 Ti with 16 GB, RTX 5090 not quite with 600 W 389 comments
Image: Nvidia
After the GeForce RTX 5090, RTX 5080 and RTX 5070, Nvidia would consider releasing the GeForce RTX 5070 Ti in the short term. New rumors clarify the key data in terms of memory expansion and TDP: there could be an additional 16 GB of VRAM, but a higher TDP. There are also some new “information items” about the RTX 5090.
GeForce RTX 5070 Ti with 16 GB of VRAM?
For the first three models mentioned, it is currently assumed that Nvidia will upgrade the 90 series to 32 GB this year after two generations with 24 GB of graphics memory, but the other models will remain at the last known level. With the GeForce RTX 5070 Ti, on the other hand, the 16 GB should correspond to the current “super level”, because the GeForce RTX 4070 Ti initially only offered 12 GB; it was only the upgrade to the RTX 4070 Ti Super at CES 2024 in January that made the 16GB jump.
As for TDP, wccftech reports an increase from the current 285 to 350 watts. kopite7kimi’s
RTX 5090 with “just under 600 watts”?
At the same time, the same account believes that it is possible that the RTX 5090 will not end up at 600 watts ex factory.
However, this information should be viewed with caution anyway, because with 600 watts ex works for the Founders Edition, custom designs could only appear with a higher TDP if they relied on two 12V 2×6 sockets, which do not only come from a small number of takes. power supplies are offered.
RTX 5080 vs. RTX 5070 Ti
But back to the RTX 5070 Ti: As for the GPU, this model is likely to use the same silicon as the RTX 5080: according to rumors, it would rely on the GB203 GPU in full configuration (-400) with 84 streaming assets. multiprocessors (SM), while the 5070 Ti of which 70, or 17 percent, could be used less. There should also be a difference in memory speed: the 5080 should use GDDR7 with 32 Gbps, while the 5070 Ti should only use fast 28 Gbps chips.
Currently, the 4080 Super and 4070 Ti Super also separate 14 shader clusters, or 17% active units, as well as a memory speed of 2 Gbps (23 to 21 Gbps).
GeForce RTX 5000 Series Key Data Overview. Model GeForce RTX 5090 GeForce RTX 5080 GeForce RTX 5070 Ti GeForce RTX 5070 GeForce RTX 5060 Blackwell Architecture GPU GB202-300 GB203-400 GB203-300 GB205 ? Active CPUs (chip) 170 (192) 84 (84) 70 (84) 50 (50) ? active shaders 21,760 10,752 8,960 6,400 ? Clock speeds? Memory 32 GB GDDR7 16 GB GDDR7 16 GB GDDR7 12 GB GDDR7? Memory interface 512-bit 256-bit 256-bit 192-bit? Memory speed 28 Gbps 32 Gbps 28 Gbps 28 Gbps ? Bandwidth 1792 GB/s 1024 GB/s 896 GB/s 672 GB/s ? TDP 600W 400W 350W / 285W? 250W? Market launch in the first quarter of 2025? All information unconfirmed! Topics: GeForce GeForce RTX 50 graphics cards Nvidia Nvidia Blackwell Source: X (Twitter)
An engineer by training, Alexandre shares his knowledge on GPU performance for gaming and creation.
