RTX 2050 and MX500 benchmarks on 3DMark

A few hours after the unexpected introduction, we see the first benchmarks of Nvidia’s new low-end GeForce RTX 2050 and GeForce MX500.

These tests are from the platform Zhihu Originated where the results are based on pre-production samples. Of the three models introduced, only two GeForce RTX 2050 and GeForce MX550 graphics are present in this revelation. It looks like the RTX 2050 and MX570 will perform similarly, at least in this 3DMark test.

GeForce RTX 2050 and GeForce MX500 graphics benchmarks

It should be noted that Nvidia has told ComputerBase that both models are actually the same (based on the Ampere GA107 chip with 2048 cores), which means that we see support for beam interception at the hardware level and DLSS (by Tensor cores). . However, RT cores may be a bit limited, but there is no specific explanation. In addition, the RTX 2050 has 4 GB of memory, while the MX570 is equipped with only 2 GB of memory. Of course, this does not have much effect on the Time Spy benchmark.

Confirmed specifications for both MX500 graphics show that their bus is limited to 64 bits and the memory is 12 Gbps. The clock speed of the MX570 is 1155 MHz and the MX550 is 1320 MHz. The MX550 is actually a reduced version of the GeForce GTX 1650 with the Turing TU117 chip. 1024 CUDA cores are preserved but the memory bus is reduced to half that is 64 bits. At the same time, memory technology has been improved in the new model from GDDR5 to GDDR6.

Test results are not great, GeForce RTX 3050 with 128-bit bus scores 4800 in Time Spy graphics. It should be noted again that the results are based on pre-production samples. In fact, these chips are supposed to be paired with next-generation processors like the Alder Lake M, which means we don’t see the ultimate performance yet. These graphics will be officially released in the spring of 2022.