China-Exclusive RTX 5090 D V2 Benchmarks Reveal 2% Slower Gaming Performance Than RTX 5090 D, And A Sharp Decline In Productivity And AI Prowess

Aug 12, 2025 at 10:48am EDT
GeForce RTX 5090 D V2 graphics card with futuristic tech background.

The new NVIDIA RTX 5090 variant just hit the shelves in China, and as per the latest benchmarks, it's as powerful as the RTX 5090 D but only in gaming and synthetic tests.

GeForce RTX 5090 D V2 Trades Blows With RTX 5090 D in Synthetic and Gaming Tests, But Shows up to 25% Performance Decline in Productivity and 10% in AI Tests

The latest benchmarks published by Expreview confirm that the new memory configuration on the newly launched GeForce RTX 5090 D V2 isn't limiting the GPU in gaming. This was obvious from the very start since 24 GB memory capacity is more than sufficient for modern titles, even at 4K, as a number of specifications are retained on the new card. Still, there must be applications where the lower VRAM capacity would impact the overall performance. The review confirms that the RTX 5090 D V2 indeed impacts the performance negatively in a bunch of productivity and AI tests.

Related Story GIGABYTE Pushes Into The AI Assistant Race With GiMATE, Powering Three New AORUS And AERO Laptops Built Around RTX 50 Series GPUs

As far as gaming and synthetic tests are concerned, there is hardly a 1-2% performance difference between the two. The RTX 5090 D V2 can be slightly slower in some instances, but overall, you should expect the same performance as the previous RTX 5090 D variant. At 4K native resolutions, multiple games confirmed almost no performance regressions, which is good news for gamers, but the 8 GB reduced VRAM capacity is impacting the productivity performance greatly.

While synthetic benchmarks reveal similar performance to the RTX 5090 D, including the popular 3DMark tests like Fire Strike, Time Spy, Port Royal, Speed Way, and Steel Nomad, in productivity applications such as Blender and V-Ray, things don't look good at all. In Blender, a performance gap of as high as 19% was seen in Junkshop, while the Monster and Classroom scenes showed a 10.5% and 15.9% performance gap, putting the RTX 5090 D V2 significantly behind the RTX 5090 D variant.

In V-Ray Benchmark, the rendering performance of RTX 5090 D V2 came out roughly 7.5% slower than RTX 5090 D in CUDA, and around 20% slower (or a 24.8% difference) in the RTX benchmark. This is quite a downgrade while the RTX 5090 D V2 was seen listing for nearly the same price as the RTX 5090 D. However, the official price is roughly $500-$550 lower than the latter when we compare the actual market prices of both (since RTX 5090 D was available for much higher than its MSRP). In AI tests, such performance drops were seen across multiple tests, including various UL Procyon AI tests, particularly the LLAMA 3.1 and LLAMA 2, which show as much as a 10% performance difference.

All these performance drops point to the lower VRAM capacity on the RTX 5090 D V2, making it unappealing for AI workloads. The RTX 5090 D already came with nerfed AI capabilities, and now with reduced VRAM capacity, this is more obvious. Perhaps, the RTX 5090 D V2 might serve pretty much well for gaming needs, but for productivity and AI, it just can't compete with the previous variant. However, if users can find it for its official MSRP of 16,499 Yuan (roughly US$2295, the same as the launch price of RTX 5090 D), it's still a decent alternative.

News Source: Expreview

About the author: Sarfraz Khan is a hardware reporter with a focus on PC components and the builder community. With years of experience writing about PC hardware and laptops, his work has been featured on several reputable technology publications. Sarfraz's hands-on experience is demonstrated through his first-person accounts of using and comparing different hardware configurations, providing practical and relatable insights for everyday users. His technical analysis is respected by peers in the enthusiast community and has been cited by specialized hardware sites such as Germany's Igor's Lab.

Follow Wccftech on Google to get more of our news coverage in your feeds.