In an intriguing development within the world of mobile graphics technology, the GeForce RTX 5090 Laptop GPU has emerged as the number one contender on PassMark’s G3D Mark performance chart for high-end video cards. With an impressive score of 28,280, it currently surpasses the RTX 4090 Laptop GPU, marking it as the top-performing mobile GPU ever recorded on the platform. It stands toe-to-toe with desktop powerhouses—positioned slightly above the AMD Radeon RX 6950 XT and just a hair below the Radeon PRO W7800.
Depending on your perspective, this feat may be a cause for celebration or concern. Die-hard Nvidia enthusiasts might rejoice as the RTX 5090 Laptop GPU proves its mettle with a significantly lower thermal design power (TDP) than its desktop counterparts. On the flip side, skeptics may point to the older desktop cards the RTX 5090 outshines and question the marginal performance improvement over its predecessor, the RTX 4090.
However, the transition from Ada Lovelace to the Blackwell architecture shines brightly for Nvidia fans. The RTX 5090 Laptop GPU boasts a hefty 24 GB of GDDR7 memory compared to the RTX 4090’s 16 GB GDDR6, alongside a boost in CUDA cores from 9,728 to an impressive 10,496. Enhanced by the latest Max-Q technologies, this new series promises greater efficiency and optimization.
Yet, curiosity lingers around why the performance margin over the RTX 4090 is a mere 2.79%. This disparity is attributed to compatibility challenges, particularly with Nvidia halting support for 32-bit frameworks. This hiccup resulted in surprisingly lackluster performances in specific tests, notably the GPU Compute test. However, a patch is reportedly in the works to ensure the RTX 50-series is more accurately represented.
Despite these road bumps, the RTX 5090’s performance in DirectX 9-12 tests showcases its formidable capabilities, leaving the RTX 4060 trailing in its wake. If this is just the beginning, Nvidia’s flagship laptop GPU commands attention, signaling an exciting future in mobile graphics technology.
