Nvidia Confirms RTX 5050 Will Use GDDR6 Memory from Samsung and SK hynix, Not GDDR7 for Launch

Recent developments have clarified that Nvidia’s upcoming RTX 5050 GPU won’t utilize the cutting-edge GDDR7 memory, opting instead for established GDDR6 modules from Samsung and SK Hynix. According to reports from Benchlife.info, Nvidia’s add-in board partners have confirmed this choice, effectively squashing any hopes for GDDR7 in the initial RTX 5050 launch. The decision to stick with GDDR6 is strategic. This memory type is well-established and widely produced, which helps reduce manufacturing costs and mitigates supply chain risks associated with newer memory technologies. Given that GDDR6 is available from multiple suppliers beyond Samsung and SK Hynix, having these two providers ensures a robust supply chain. This flexibility allows Nvidia to switch between suppliers based on availability and demand, ensuring smooth production. While the exact performance details of the GDDR6 modules for the RTX 5050 have not been revealed, GDDR6 offers a range of speed options. Current speeds range from 12 Gbps to 20 Gbps, with more recent cards using 16 Gbps, 18 Gbps, and 20 Gbps versions. If Nvidia opts for the higher end of this spectrum, such as 18 Gbps, the RTX 5050 would achieve 288 GB/s of memory bandwidth. For comparison, the RTX 5060 is expected to offer nearly double this, with 448 GB/s. Early rumors indicate that the RTX 5050 will come equipped with 8 GB of GDDR6 memory and consume 130W of power. The entry-level Blackwell GPU is expected to feature a fully enabled GB207 die, including 2,560 CUDA cores, 20 streaming multiprocessors (SMs), and a 128-bit memory interface. With 18 Gbps GDDR6, this configuration would deliver the aforementioned 288 GB/s of memory bandwidth. Despite the initial decision to use GDDR6, history suggests that Nvidia may introduce a GDDR7 version of the RTX 5050 in the future. The company has a precedent of releasing multiple variants of its xx50 class cards, each with different memory types. For instance, the GTX 1650, launched in 2019 with 8 Gbps GDDR5 memory, saw an upgrade to 12 Gbps GDDR6 less than a year later. Additionally, different revisions like the TU106 and TU116 versions further diversified the GTX 1650 lineup, offering the same specs as the GDDR6 variant. Moreover, the mobile version of the RTX 5050 is said to sport GDDR7 memory, hinting at the possibility of a similar upgrade for the desktop model. Should Nvidia follow through, the RTX 5050 Super variant could potentially feature 12 GB of 28 Gbps GDDR7, aligning with the company's rumored plans for a Super refresh of its Blackwell series. This refresh is expected to update at least some of the existing GPUs with 3 GB GDDR7 integrated circuits (ICs). For those interested in staying abreast of the latest developments in the tech world, make sure to follow Tom's Hardware on Google News. By clicking the Follow button, you can receive their up-to-date news, analysis, and reviews directly in your feeds.