On Friday, Nvidia launched a trio of new laptop GPUs that it says will start to be featured in laptops starting from spring 2022. It announced the GeForce RTX 2050, MX570, and MX550 GPUs, aimed at providing “more GeForce laptop choices for gamers and creators.” It is great to get more choice, but Nvidia’s post was quite useless, with no specs discussed or performance targets hinted at.
As tech sites scrambled to cover the news ahead of the weekend, the blog post and product pages provided by Nvidia were woefully light on detail. It wasn’t until Saturday that some key questions were cleared up. For example, AnandTech Editor-in-Chief, Ryan Smith, received confirmation that both the GeForce RTX 2050 and MX570 are based on the A107 Ampere GPU (Samsung 8nm). Meanwhile, the MX550 is going to be TU117 Turing GPU based (TSMC 12nm). The RTX 2050 specs have also now been added to the official 20-series gaming laptops comparison tables on the GeForce site.
In the table below I have tabulated the specs of the new GeForce RTX 2050, MX570, and MX550 and thrown in some existing relations to contextualise their standing.
| RTX 3060 Laptop GPU | RTX 3050 Laptop GPU | RTX 2050 Laptop GPU | MX570 | MX550 | MX450 | |
| CUDA Cores | 3,840 | 2,048 | 2,048 | >1,024? | 1,024? | 896 |
| ROPs | 48 | 32 | 32? | 32? | 32? | 32 |
| Boost Clock | 1,283 – 1,703MHz | 1,057 – 1,740MHz | 1,155 – 1,477MHz | ? | ? | 1,395 – 1,575MHz |
| Memory Clock | 14Gbps GDDR6 | 14Gbps GDDR6 | 14Gbps GDDR6 | 12Gbps GDDR6? | 12Gbps GDDR6? | 10Gbps GDDR6 |
| Memory Bus Width | 192-bit | 128-bit | 64-bit | 64-bit | 64-bit | 64-bit |
| VRAM | 6GB | 4GB | 4GB | ? | ? | 2GB |
| TDP Range | 60 – 115W | 35 – 80W | 30 – 45W | <30W? | <30W? | <30W |
| GPU | GA106 | GA107 | GA107 | GA107 | TU117 | TU117 |
| Architecture | Ampere | Ampere | Ampere | Ampere | Turing | Turing |
| Manufacturing Process | Samsung 8nm | Samsung 8nm | Samsung 8nm | Samsung 8nm | TSMC 12nm | TSMC 12nm |
| Launch Date | 01/26/2021 | 05/11/2021 | Spring 2022 | Spring 2022 | Spring 2022 | August 2020 |
Table compiled from Nvidia, AnandTech, and NotebookCheck data
Given its Ampere pedigree, it is quite strange how Nvidia has introduced the RTX 2050 for laptops. Of course, there is already an RTX 3050/Ti, and it traditionally releases ##40 GPUs as part of the GT line, so this is what it came up with as a compromise. You can see from the table the RTX 2050 for laptops is a boost clock/power-constrained RTX 3050 using half the bus width (64-bit).
Another interesting thing to observe is that the MX570 and MX550 are likely to have the same cores/memory and power configuration but vary only by architecture. There are reports that the MX570 will have the advantage of supporting DLSS, but this isn’t mentioned by the official product page.
There have already been some 3DMark Timespy scores for the new mobile GPUs spotted by HXL on Twitter. The non-final test systems achieved the following scores; “RTX 2050/MX570 GPU Score ~3,300 / MX 550 GPU Score ~2,500.” For reference, the average Nvidia GeForce GTX 1650 Mobile scores 3,571 points, and the average GeForce GTX 1050 Ti Mobile scores 2,451 points, according to NotebookCheck.
