Rumor has it that the Nvidia GeForce RTX 3080 TI could offer 21 TFLOPs of raw performance, 50% more than the current Nvidia GeForce RTX 2080 Ti. Twitter tipster @CatCyberPunk shared an alleged datasheet for the incoming Ampere GPU, revealing that the Nvidia GeForce RTX 3080 TI will have 5,376 CUDA cores operating at up to 2,2GHz, an improvement over current Turing GPUs that barely reach 2.1 GHz. However, as Wccftech pointed out, this doesn't match up, because with 21 TFLOPs of computing power, the official boost clock would be 1953 MHz. The as-yet-unverified leak also claims that the Nvidia GeForce RTX 3080 Ti will have 5,376 CUDA cores, 12 GB of GDDR6 memory running at 18 bit/s, giving it a total bandwidth of 863 GB/s and, despite its 7nm architecture, a 320W TDP
Amps vs Big Navi
Although the validity of the leak has not been confirmed, if the datasheet is legitimate, it could herald bad news for the upcoming AMD Big Navi GPU. As Notebookcheck noted, early leaks indicated that Big Navi only provides 10-15% better performance than the GeForce RTX 2080 Ti, so if the GeForce RTX 3080 Ti's spec list is bona fide, AMD could be considerably disadvantaged when it's all about high-end GPU performance. The Nvidia GeForce RTX 3080 Ti should debut alongside the RTX 3080, which will carry 10GB of VRAM, and the higher-end RTX Titan which is said to have 24GB of VRAM. This latest leak from Ampère comes a few days after benchmarks were posted online comparing the Nvidia GeForce RTX 3080 to the RTX 2080 Ti. Some rumors have even suggested the existence of an RTX 3090, but we doubt that such a GPU exists. According to leaked benchmarks, which are popular 3D Mark Time Spy tests, an unknown GPU from Nvidia scores an impressive 18,257. That's a huge 30.98% jump compared to the RTX 2080 Ti, which is currently Nvidia's most powerful consumer gaming graphics card. We hope we don't have to wait too long to find out what Nvidia's next-gen GPU plans are.