NVIDIA today officially announced the GeForce RTX 30506GB as its new entry-level discrete GPU. The RTX 30506GB is significantly different from the RTX 3050 that the company launched as a mid-range product back in January 2022.
The RTX3050 was originally built on the GA106 chip in an 8nm process, with 2560 CUDA cores, 80 Tensor cores, 20 RT cores, 80 TMUs and 32 ROPs, as well as 8GB14Gb on a 128-bit bus psGDDR6 video memory; these specifications also match the maximum core configuration of the smaller GA107 chip, so the company launched the GA107-based RTX3050 in late 2022 with unchanged specifications but slightly improved energy efficiency due to the switch to a smaller chip. The new RTX 30606GB is based on the same GA107 chip, but with significant changes.
The most obvious change in the RTX30506GB version is the video memory. The new SKU is equipped with 6GB 14Gbps GDDR6, achieving 168GB/s memory bandwidth through a narrower 96-bit memory bus. Not only that, the GPU has also been greatly reduced, with only 16 SMs instead of 20 on the original RTX3050. This equates to 2048 CUDA cores, 64 Tensor cores, 16 RT cores, 64 TMUs, and a constant 32 ROPs. The GPU is clocked at a lower 1470MHz, compared to the original RTX3050's 1777MHz.
The highlight of this SKU is that its total graphics power consumption (TGP) is only 70W, which means that the graphics card can completely eliminate the need for external power and rely entirely on PCIe slot power. Nvidia hasn't listed an MSRP for this product yet, but last we heard it was supposed to cost $179 and compete with the Intel ArcA580.