NVIDIA is coping with the chip crisis and expanding its range of mobile graphics chips. In 2022, the GeForce RTX 2050 will appear on the market but based on the Ampere architecture.
GeForce RTX 2050 is almost a twin of the 3050 model
Graphics cards are constantly scarce on the market and gamers either have to decide to buy at inflated prices or turn their eyes to notebooks. And they probably do it more and more often, and NVIDIA intends to further expand its offer in this segment. At the same time introduces a bit of confusion. Recently, the GeForce RTX 3050 based on the Ampere architecture has appeared on the market. It is a slightly weaker model than the RTX 3060, but it supports ray-tracing and offers quite good performance that allows you to play at 1080p resolution. Lower in the hierarchy is the GeForce GTX 1650 Ti, which is based on the Turing architecture (the same as the RTX 20 series) but without ray-tracing support. There is a big performance gap between the RTX 3050 and the GTX 1650 Ti, and that’s where the RTX 2050 will fit in.
The name might suggest that this is another card based on the older Turing architecture, but it turns out to be the Ampere GA107 chip, the same one found in the RTX 3050, equipped with 2048 CUDA cores and clocked from 1155 to 1477 MHz in Boost mode . Power consumption is to range from 30-45 W, which means that it will be a system rather installed in cheaper notebooks. Its main limitation is memory bandwidth. We have 4 GB of GDDR6 memory at our disposal, but working only on a 64 bit wide bus (128 bit in RTX 3050), which gives a bandwidth of 112 GB / s. This is a significant limitation and a bottleneck that will be visible mainly at higher resolutions.
Nevertheless, if the computers with RTX 2050 are well priced, they can be quite an interesting alternative. Like the GeForce MX570 and MX550 chips, which will hit the market at the same time, which is next spring. Unfortunately, we know even less about them and it is difficult to guess what performance they can offer.