Chinese GeForce RTX 4090 D sips less power, can’t overclock

The 'D' doesn't stand for downclocked, believe it or not.

GeForce RTX 4090 Dragon - background by mathew schwartz unsplash.
Background by Mathew Schwartz via Unsplash

Nvidia’s GeForce RTX 4090 D is on track to launch in Asia markets before the end of January 2024. The ‘for gaming only’ GPU is set to replace the original RTX 4090 due to US regulations imposed on China. The card will reportedly feature a lower TGP and will not support overclocking.

At least that’s according to the latest follow up from Benchlife (via MEGAsizeGPU.) The RTX 4090 Dragon is said to feature a cut down version of the AD102-300-A1 GPU, dubbed AD102-250, though total CUDA cores remain a mystery.

Its reduced 425W TGP (from 450W) indicates there’ll be a small yet significant CUDA cores and transistor count reduction. For comparison, The RTX 4090 FE features 76 billion transistors and 16,384 CUDA cores, mated to 24GB of GDDR6X VRAM. Video memory will most likely remain unscathed.

Moving on, the source suggests a higher base clock at 2,280MHz (previously 2,235MHz,) and a boost clock of 2,520MHz. The latter equals the same as the original boost clock. That’s about it regarding confirmed specifications for the restricted card.

From what we’ve gathered last month, it remains evident that Nvidia intends to stay close to the original specifications as possible, while still meeting US export guidelines to ensure the survival of the RTX 4090 in China. Unfortunately, the performance reduction doesn’t mean a price decrease, as Nvidia plans to launch at the same MSRP of ¥12,999 (£1,451.) Sucks to be a PC gamer in China.

For now, the source suggests samples of the AD102-250 GPU are currently being doled out to Nvidia AIB Partners for proofing and testing this week. There’s still plenty of time for even more leaks over the coming weeks.