The tech giant unveiled the RTX 4090D on its Chinese website and admitted it was a lot weaker than the model it launched in late 2022.

The card has just 14,592 CUDA cores compared to 16,384 on the versions sold elsewhere. Nvidia said the card's tensor core count was also cut from 512 to 456 on the 4090D.

Otherwise, the card is pretty much the same, with top speeds of 2.52 GHz, 24 GB of GDDR6x memory, and a massive 384-bit memory bus.

The RTX 4090 was the only gaming card banned from sale in China after the Biden Administration slapped on the toughest export rules ever in October.

The problem was the card was too powerful for consumer cards, with a total processing performance (TPP) of more than 4,800. That number is worked out by doubling the max number of dense tera-operations per second - floating point or integer - and multiplying by the bit length of the operation.

The original 4090 had a TPP of 5,285, which meant Nvidia needed a US government license to sell the hot gaming card in China. Note that consumer cards don't have to meet the performance density limit that stops the sale of much weaker data centre cards like the Nvidia L4. Cutting performance by 10.94 per cent was enough to get the card under the limits that make the USA's Bureau of Industry and Security (BIS) look at an export license.

Nvidia says that the 4090D can be overclocked by users, which means they can get back some of the performance they lost by the lower core count.

 "In 4K gaming with ray tracing and deep-learning super sampling (DLSS), the GeForce RTX 4090D is about five per cent slower than the GeForce RTX 4090, and it works like every other GeForce GPU, which users can overclock," an Nvidia spokesperson said.