New rumors are coming related to NVIDIA Ampere and the flagship GPU for the gaming segment, the Nvidia GeForce RTX 3080 Ti. First of all, NVIDIA Ampere might not eventually come to TSMC’s high-performance 7nm (HPC) manufacturing process, but could instead employ Samsung’s 10nm (8LPP) manufacturing process.
On the other hand, on Twitter KittyCorgi revealed the existence of at least five Ampere graphics cards, with two low-end, one mid-range, one high-end and one top-of-the-range model.
As we can see from the above chart, the most basic model will have GA107 silicon with 1280 CUDA Cores along with 4 GB of GDDR6 memory and 128-bit memory interface. Due to the improvements in architecture, we could say that it comes to replace the GeForce GTX 1660. With GA106 silicon, we have 1920 CUDA Cores, 6GB of GDDR6 memory, and a 192-bit memory interface. It will replace the GeForce RTX 2060.
Going into high-performance graphics cards, GA104 silicon appears with 3072 CUDA Cores equipped with 8 GB GDDR6 and 256 bits. This model could replace the GeForce RTX 2070 and would be 5% less powerful than the GeForce RTX 2080. With the GA103 silicon, we have 3840 CUDA Cores together with 10 GB GDDR6 and 320-bit interface, this would be the replacement of the GeForce RTX 2080 Ti being 10% more powerful than the GeForce RTX 2080 Ti.
Finally, the GA102 would be the top of the range GPU Core has 84 SM units with a total of 5,376 CUDA cores, 12 GB GDDR6 and 384 bits interface. This would be the GeForce RTX 3080 Ti and would offer 40% more performance than the GeForce RTX 2080 Ti. These numbers would be very conservative, since the full Ampere silicon, the GA100 is capable of offering a total of 8192 CUDA Cores with 1024 Tensor Cores, 130 RT Core and a TDP of 300W, although this model has been manufactured by TSMC @ 7nm +.
NVIDIA Ampere GPU ‘GeForce RTX 3080 Ti, RTX 3080, RTX 3070’ Alleged Specifications:
|Graphics Card Name||NVIDIA GeForce RTX 2070||NVIDIA GeForce RTX 2080||NVIDIA GeForce RTX 3070||NVIDIA GeForce RTX 3080||NVIDIA GeForce RTX 2080 Ti||NVIDIA GeForce RTX 3080 Ti|
|GPU Architecture||Turing TU106||Turing TU104||Ampere GA104||Ampere GA103||Turing TU102||Ampere GA102|
|Process||12nm FinFET NVIDIA (TSMC)||12nm FinFET NVIDIA (TSMC)||Samsung 10nm (8LPP)?||Samsung 10nm (8LPP)?||12nm FinFET NVIDIA (TSMC)||Samsung 10nm (8LPP)?|
|CUDA Cores||2304 CCs||3072 CCs||3072 CCs||3840 CCs||4608 CCs||5376 CCs|
|Memory||8 GB GDDR6||8 GB GDDR6||8/16 GB GDDR6||10/20 GB GDDR6||11 GB GDDR6||12 GB GDDR6|
|Memory Speed||14 Gbps||14 Gbps||16 Gbps?||16 Gbps?||14 Gbps||16 Gbps?|
|Memory Bandwidth||448 GB/s||448 GB/s||512 GB/s||640 GB/s||640 GB/s||768 GB/s|
We will have to wait for more information to confirm who is behind the manufacturing of these silicons. Otherwise, all of these GPUs will have RayTracing support, and the GeForce RTX 3080 Ti would be the only model with Nvidia SLI support.