Update 12/17/2021 10:19 PT:
Nvidia has confirmed to German news outlet Computerbase the architectures behind the GeForce RTX 2050, MX570 and MX550 graphics cards. We've amended the story to reflect the changes.
Original Article
Nvidia slipped out a quiet announcement of the arrival of three new mobile graphics cards in the spring of 2022. The GeForce RTX 2050 will join the ranks of the GeForce MX570 and MX550 in powering a new wave of laptops for gamers and creators.
The GeForce RTX 2050 and GeForce MX570 are based on the GA107 (Ampere) silicon, the same silicon that powers the GeForce RTX 3050 and RTX 3050 Ti Mobile. The GeForce MX550, on the other hand, is still based on the TU117 (Turing) silicon from the current GeForce MX450.
In regards to featureset, the GeForce MX570 will support Nvidia DLSS and ray tracing in a "limited" fashion. Although the GeForce MX550 has the Tensor and RT cores bake in it, the Turing-based graphics card do not support Nvidia DLSS or ray tracing. As a result, it doesn't support Nvidia Broadcast unlike the GeForce RTX 2050 and MX570. However, the trio of upcoming GeForce graphics card are compatible with the Resizable BAR and Optimus technologies, but not Advanced Optimus.
GeForce RTX 2050 Specifications
| Header Cell - Column 0 | GeForce RTX 2080 Super | GeForce RTX 2080 | GeForce RTX 2070 Super | GeForce RTX 2070 | GeForce RTX 2060 | GeForce RTX 2050 |
|---|---|---|---|---|---|---|
| Nvidia CUDA Cores | 3,072 | 2,944 | 2,560 | 2,304 | 1,920 | 2,048 |
| Boost Clock (MHz) | 1,080 - 1,560 MHz | 1,095 - 1,590 MHz | 1,155 - 1,380 MHz | 1,125 - 1,455 MHz | 1,185 - 1,560 MHz | 1,155 - 1,477 MHz |
| Graphics Subsystem Power (W) | 80 - 150+ W | 80 - 150+ W | 80 - 115 W | 80 - 115 W | 65 - 115 W | 30 - 45 W |
| Memory Speed | Up to 14 Gbps | Up to 14 Gbps | Up to 14 Gbps | Up to 14 Gbps | Up to 14 Gbps | Up to 14 Gbps |
| Standard Memory Configuration | 8 GB GDDR6 | 8 GB GDDR6 | 8 GB GDDR6 | 8 GB GDDR6 | 6 GB GDDR6 | 4 GB GDDR6 |
| Memory Interface Width | 256-bit | 256-bit | 256-bit | 256-bit | 192-bit | 64-bit |
| Memory Bandwidth (GB/sec) | Up to 448 GB/s | Up to 448 GB/s | Up to 448 GB/s | Up to 448 GB/s | Up to 336 GB/s | Up to 112 GB/s |
Nvidia listed the GeForce RTX 2050 with 2,048 CUDA cores, around 7% more than the GeForce RTX 2060. However, the GeForce RTX 2050 adheres to a more restricted TDP so it'll come with lower clock speeds. The GeForce RTX 2060 should still be faster. As usual, Nvidia gives its partners some thermal headroom, meaning the power limit on the GeForce RTx 2050 will vary from one product to another. In any event, the graphics card is rated between 30W to 45W. Therefore, the boost clock will vary as well, and we'll find the GeForce RTX 2050 boosting between 1,155 MHz and 1,477 MHz.
In terms of the memory configuration, the GeForce RTX 2050 has 4GB of GDDR6 memory clocked at 14 Gbps. Bound by a 64-bit memory interface, the GeForce RTX 2050 can deliver a maximum memory bandwidth up to 112 GBps.
Meanwhile, the GeForce MX550 will be the direct replacement for the GeForce MX450, whereas the GeForce MX570 appears to be a new SKU. Nvidia didn't delve into the intricacies of the GeForce MX570 and MX550. The product pages confirm support for PCIe 4.0 and GDDR6 memory.
The GeForce MX570 and MX550's other specifications remain a mystery. The chipmaker only stated that the duo sport "more power-efficient CUDA cores and with faster memory speeds than previous MX GPUs." In addition, Nvidia claims that the GeForce MX570 and MX550 will offer a leap of performance and efficiency in photo and video-editing tasks, such as Adobe Lightroom and Premiere Pro, as well as improved gaming in comparison to integrated graphics.
There's a budget GeForce GPU selling in China that not even Nvidia knew it made — RTX 4010 turns out to be a modified RTX A400 workstation GPU
US to patch loopholes that allow China to buy banned AI GPUs from other countries — new regulations include national quotas on GPU exports and a global licensing system
