The tech world is abuzz with NVIDIA’s upcoming release tailored specifically for the Chinese market: the GeForce RTX 5090D. While this GPU carries forward the legacy of its predecessor, the RTX 4090D, NVIDIA is introducing some noteworthy restrictions aimed at altering its use in AI and crypto-mining applications.
A significant tweak in the RTX 5090D is its “performance cut-off” mechanism. After just three seconds of engaging in AI or cryptocurrency-related tasks, the GPU’s performance is curtailed, likely through reduced clock speeds or other onboard software throttling methods. This restriction also extends to power limits, meaning the 5090D may not deliver optimal performance in intense computing workloads.
Moreover, one of the major changes is the prohibition of using the RTX 5090D in multi-GPU setups. Previously, enthusiasts managed to employ the RTX 4090D in such configurations, drawing gamer-level performance suitable for AI clusters. These setups will no longer be viable as NVIDIA enforces stringent blockades, even affecting Linux users.
Despite these limitations, gamers worried about their experience can breathe easy. These restrictions are designed specifically to curb professional AI and mining applications, leaving gaming performance unaffected.
Highlighting the specifications, the RTX 5090D is built on the GB202 GPU die and features an impressive 21,760 CUDA cores and a formidable 32 GB of GDDR7 memory running at 28 Gbps, granting a bandwidth of 1792 GB/s. Several AIB models are already surfacing, indicating readiness in the market.
As part of the larger RTX 50 series, NVIDIA plans to roll out varied models such as the RTX 5090, RTX 5080, 5070 Ti, and others, each offering unique specifications tailored to different consumer needs. With tentative launch dates ranging from January to April 2025, the anticipation is building up for what these new GPU offerings will bring to the table, both in China and potentially worldwide.
