Select Language:
In a significant development for gaming hardware enthusiasts, NVIDIA is reportedly set to unveil the RTX 5090 D v2 graphics card this August. The news stems from notable leaker Kopite7kimi and is complemented by information from another source, Benchlife.
As the market evolves, NVIDIA’s plans have been influenced by recent export restrictions, which initially barred the RTX 5090 from being sold in China due to its excessive computational power. To address these regulations, NVIDIA created a specially designed RTX 5090 D model for the Chinese market. However, early user feedback pointed to performance and stability issues, and evolving export standards quickly rendered the RTX 5090 D nearly obsolete.
During the ensuing months, Chinese gamers only had access to the RTX 5080 through official channels, leaving the RTX 5090 and its predecessor, the RTX 4090, available primarily through third-party vendors without official warranty support.
To address the needs of the gaming community, NVIDIA is taking measures to introduce compliance-friendly high-end models. The RTX 5090 D v2 will maintain the core specifications of its predecessors, utilizing the GB202-240 core and housing 21,760 CUDA cores. A notable update is the transition to GDDR7 memory, featuring 24GB capacity with a 384-bit bus width and 28Gbps frequency. However, total memory bandwidth will decrease from 1,790 GB/s to 1,342.5 GB/s.
Recent analyses suggest that the changes in the RTX 5090 D v2 are relatively minor compared to the original RTX 5090 D, primarily involving a reduction from 16 to 12 GDDR7 memory chips. This change should still yield a performance level surpassing that of the older RTX 4090.
Concerns regarding memory capacity and bandwidth seem assuaged; even with the modifications, the RTX 5090 D v2 is unlikely to detract from gaming performance, particularly at 4K resolution, where games rarely demand beyond 24GB of memory.
Moreover, the RTX 5090 D v2 is expected to maintain a thermal design power (TDP) rating of 575W. The reduced memory count may result in a lower overall power draw, allowing more power to be allocated to the core and potentially enhancing gaming performance. However, concrete results will await further testing.
Additionally, the supply chain for the RTX 50 series is diversifying, with Micron now listed as a supplier alongside Samsung and Hynix. This move suggests that upcoming models, including the RTX 5090 D v2 and the future RTX 50 SUPER series, may feature variations in memory performance due to the differing frequencies of GDDR7 memory chips.
Notably, the SUPER series may offer enhanced memory specifications, ranging from 2GB to 3GB per chip, enabling models like the RTX 5070 Ti SUPER and RTX 5080 SUPER to compete in memory capacity with the RTX 5090 D v2. This could be excellent news for consumers focused on specific applications, such as AI, video editing, rendering, and modeling.
Add us on ChatGPT
Add us on Perplexity
