As anticipated, NVIDIA has officially introduced the RTX 4090D today. Positioned as a direct substitute for the RTX 4090 card, which encountered import restrictions in China—a crucial gaming market for NVIDIA—the company had to make strategic choices. Instead of introducing the rumored RTX 4080 SUPER as the latest flagship, NVIDIA chose to adapt the current RTX 4090 SKU to adhere to U.S. government restrictions on GPU exports to China.
In contrast to previous speculations, NVIDIA has officially acknowledged modifications to the specifications of the RTX 4090. The RTX 4090D variant now features a reduced core count, dropping from 16,384 to 14,592, indicating a significant alteration. Furthermore, the Tensor core count has been adjusted to 456 cores, down from the original 512.
In terms of memory specifications, the card remains consistent with 24GB of GDDR6X memory, operating at a clock speed of 21 Gbps and utilizing a 384-bit interface. Notably, there is no change in the memory specifications. However, a significant adjustment is observed in the default Thermal Design Power (TDP), now set at 425W, signaling a 25W reduction compared to the original model.
As board partners reveal their designs, it’s clear that the visual aspects closely mirror the original RTX 4090 models. No surprises are found, and the majority of cards maintain an identical appearance to the original designs. The official price remains unchanged at 12,999 RMB. However, what has changed is the absence of a factory-overclocking option, as none of the cards announced today seem to offer such a feature.