Nvidia is preparing to launch, earlier than initially expected, its GeForce RTX 4090 “D”. Reserved for the Chinese market, this new version of the high-end card from the chameleon manufacturer has the main objective of circumventing the American embargo imposed on certain GPUs distributed in China.
RTX 4090 “D”. Does this name mean nothing to you? This is normal, and a priori, you will never hear about it in Europe. Mentioned by a first rumor last month, this version of the GeForce RTX 4090 is designed specifically for the Chinese market. In this case, it aims to replace the RTX 4090 that we know, targeted by the American embargo imposed on certain graphics cards distributed in China.
For context, the United States now prohibits the sale of the most powerful models in China, in order to curb Chinese ambitions in terms of artificial intelligence. Without these supercharged GPUs, training the best AI models takes much longer. By preventing China from acquiring them in large volumes, Washington is therefore ensuring that it slows down the economic development of its rival in this highly strategic market… especially since Chinese brands are, in the immediate future, far from being able to launch GPUs as advanced as those from Nvidia.
9 /10
A modified RTX 4090… but far from being scuttled
As it stands, this “D” version of the RTX 4090 is therefore slightly less efficient than the “D” version.classic”, including the TPP index ( Total Processing Performance) exceeds the maximum 4800 points authorized by the American authorities to market a GPU on the Chinese market. We also learned this week that this Chinese iteration of the chip would finally be ready for launch on December 28. Initially, the latter was to be marketed in 2024. It would therefore have a head start.
In detail, the RTX 4090 “D” should use a castrated version of the AD102 GPU. It would thus be limited to a lower number of cores, so as to fall below the maximum performance threshold mentioned above. On the other hand, Nvidia would bet on a slightly increased base frequency (+ 45 MHz) for the GPU of this RTX 4090 “D”. In boost, the latter would however remain clocked at 2520 MHz, i.e. as much as the classic RTX 4090.
According to information collected byWWCFTech, we would also find the same 24 GB of GDDR6X, the same memory bandwidth (more than 1 TB/s) and the same TGP (around 400 W). Same approach in terms of pricing, since Nvidia would maintain a price of 12,999 yuan for this “new» model… or a little less than 1700 euros excluding taxes.