NVIDIA GeForce RTX 5090 and RTX 5090D: Identical GPU configurations, adjusted specifications for efficiency

NVIDIA is apparently planning to launch a new flagship graphics card, the GeForce RTX 5090, and a special variant for the Chinese market, the RTX 5090D. Recent leaks suggest that both models could be based on the same GPU configuration and identical clock rates. This raises questions about the differences between the two models, particularly with regard to the adaptations to US export regulations.

The RTX 5090D first came to light through leaked packaging materials. According to reports, it is being developed to comply with the export restrictions of the United States. These regulations concern the so-called TPP value (Thermal Power Profile), which regulates a certain power consumption of hardware. Similar measures have already been implemented for the RTX 4090D, which was introduced in response to the export ban on the RTX 4090 in China. In the case of the RTX 4090D, NVIDIA relied on a slimmed-down version of the original RTX 4090, in which technical specifications were reduced to meet regulatory requirements. For the RTX 5090D, however, it is believed that NVIDIA will instead use hardware-based restrictions that keep performance within the required limits without changing the actual configuration of the GPU.

According to a member of the chipshell forum Panzerlied, which often publishes early information on new hardware products, the RTX 5090 and RTX 5090D are said to have the same GPU specifications. This presumably includes a clock speed of 2.9 GHz, which is already speculated for the RTX 5090. Both models could have up to 21,760 CUDA cores and 32 GB GDDR7 memory. Differences could instead be limited to regulatory measures that affect power consumption. For example, the RTX 5090D could use the same GPU as the RTX 5090, but be labeled with a different SKU (Stock Keeping Unit) to technically and legally differentiate it from the standard variant. NVIDIA could use a hardware-based lock here to ensure that the TPP requirements are met. This method has already been used in previous models, such as the introduction of the LHR (Low Hash Rate) lock to restrict mining usage.

NVIDIA is expected to release the RTX 5090D in January 2025 to coincide with the global launch of the RTX 5090 and RTX 5080. There is no confirmed information regarding pricing yet, but it is assumed that the costs for both models will be at a similar level, as the technical features could be almost identical. The development of the RTX 5090D shows once again how much export regulations can influence the hardware industry. For NVIDIA, this means additional development work to meet the requirements and at the same time offer competitive products for the Chinese market. However, the fact that the technical basis of the RTX 5090D and RTX 5090 is likely to remain the same could speed up the market launch and minimize production costs.

Although the exact performance differences and limitations of the RTX 5090D compared to the standard variant are still unclear, it is clear that NVIDIA is looking for a way to meet regulatory requirements without sacrificing performance. This could ultimately have an impact on future products if similar requirements remain in place. This approach could lead to an increased separation of global and regional hardware products, depending on the respective regulatory framework.

Source: Chiphell forum Panzerlied