The specs of the upcoming graphics cards NVIDIA GeForce RTX 3050 Ti e GeForce RTX 3050 would have leaked over the past few hours along with their first benchmarks. They would have been spotted in the database Geekbench and tested on non-released laptops with Intel Tiger Lake-H processors. The RTX 3050 series would include both Ti and non-Ti variants: The two GPUs that have been identified are laptop variants, however it is interesting to note that the latter feature the same basic specs as the desktop variants.
Coming to the specs, NVIDIA GeForce RTX 3050 Ti has 20 SM o 2560 CUDA Core. This should be the full-fat GA107 GPU configuration. The GPU ran at a maximum clock of 1,03 GHz and also features 4 GB of GDDR6 memory on a 128-bit bus interface. It is likely that NVIDIA has 12 Gbps memory modules, so we can expect a total bandwidth of 192 GB / s. The GPU is also expected to have a TGP of less than 100 W. The NVIDIA GeForce RTX 3050 variant features instead 16 SM o 2048 CUDA cores, with the GPU running at a maximum frequency of 1,06 GHz. These are however given from the laptop optimized versions and we can expect some performance increase in the desktop versions.
In terms of performance, in OpenCL, the NVIDIA GeForce RTX 3050 Ti is as fast as the Radeon RX 5600 XT and GeForce GTX 1080 Ti. The non-Ti variant, on the other hand, is as fast as a GTX 1080 and is also slightly faster than the GeForce GTX 1650 SUPER, which is the one it replaces at launch. Also, these are just performance OpenCL and the Ampere architecture offers a huge performance boost in the OpenCL and Vulkan APIs, so the average performance in DirectX titles would clearly be slightly lower. In addition to that, did you know that a possible release date for the RTX 3080 Ti would have leaked in the past few days?
