Almost four months after the announcement of the GeForce RTX 2050, MX570 and MX550 mobile GPUs, come on Geekbench the first ever test of the GeForce MX570, flagship of the MX line. The benchmark, based on computational capacity in OpenCL of the new NVIDIA proposal, it looks quite interesting: with a result of 44638 pointsthe MX570 is able to leave behind the GTX 1650 Ti mobile.
Underpinning NVIDIA’s low-end mobile GPU is a processor Intel Core i5-1235U Alder Lake together with 16 GB of DDR4 memory. All stuffed into a laptop from HP called Zhan 66 Pro 14a model destined for the Chinese professional market.
In addition to detaching the of about 1000 pointsthe MX570 is positioned behind the GeForce RTX 2050 a little more than 1300 points. On the other hand, RTX 2050 and MX570 are based on a GPU Ampere GA107 combined with memories GDDR6 and support for the PCI Express 4.0.
In the case of the MX570, however, the video memory stops at only 2GB where the RTX 2050 reaches 4GB. This solution is also equipped with a ray tracing unit and Tensor Core, although we do not expect great performance.
NVIDIA hasn’t released official technical specifications, but in the benchmark we talk about 16 Compute Unit which should be equivalent to NVIDIA’s Streaming Multiprocessor (SM). Ampere architecture counts 128 CUDA cores per SM, so we’re talking about 2048 CUDA core, such as the RTX 2050. Operating frequencies and quantity of combined VRAM should be at the origin of the performance difference.
According to NVIDIA’s claims, the GeForce MX570 would be three times faster than the integrated Iris Xe GPU of Intel CPUs. All that remains is to wait for the arrival of the new laptops with the GPU in question, scheduled for this spring, to see the real results offered by the new MX570.