GeForce RTX 5050 to use faster 20 Gbps GDDR6 memory, same as RDNA 4

Although NVIDIA hasn't formally announced a new budget desktop gaming GPU, several sources point to the company launching the GeForce RTX 5050 in July. The company's 50-Class GPUs are considered entry-level, with the RTX 5050 set to be built using the GB207-300 chip with 2560 CUDA Cores, which is 33% fewer than the RTX 5060's and 16% fewer than the RTX 4060's CUDA Core count.

As we learned in prior leaks on the upcoming RTX 5050, the GPU will be the first graphics card in the GeForce RTX 50 Series line-up to not feature fast GDDR7 memory. Instead, its memory configuration will include 8GB of GDDR6. However, according to a new post by MEGAsizeGPU over social media, the GDDR6 in the GeForce RTX 5050 will comprise 20 Gbps memory modules.

This is the same faster GDDR6 memory found in AMD's Radeon RX 9000 Series, which means it will be quicker than the 8GB of VRAM found in the GeForce RTX 4060.

The RTX 5050's overall memory bandwidth on a 128-bit interface is set to hit around 320 GB/sec, compared to 272 GB/sec on the RTX 4060. This is on par with the Radeon RX 9060 XT.

MEGAsizeGPU also notes that NVIDIA is changing the design with a 3+2 phase VRM design for the GPU and memory compared to the single-phase memory design found in cards like the GeForce RTX 4060. As we never got a GeForce RTX 4050 from NVIDIA, it'll be interesting to see how much faster the GeForce RTX 5050 is compared to the last 50-Class desktop GPU - the GeForce RTX 3050.

There's no word on an exact launch date or pricing for the RTX 5050, so stay tuned.