NVIDIA GeForce RTX 4090 Laptop GPU vs NVIDIA GeForce MX250

NVIDIA GeForce RTX 4090 Laptop GPU

The NVIDIA GeForce RTX 4090 Laptop GPU (Codename GN21-X11) is a high-end laptop GPU. It is based on the AD103 chip as the desktop RTX 4080 and use the Ada Lovelace architecture. The AD103 chip offers a maximum of 10.752 shaders from which inly 9728 are used (e.g., the RTX 4080 Desktop uses 9.728) and a 256 Bit memory bus. The RTX 4090 mobile uses 16 GB GDDR6 dedicated graphics memory with a clock speed of 20 Gbps (effective). The TGP (Total Graphics Power) can be configured between 80 and 150 Watt (+ 15 Watt Dynamic Boost from the CPU) leading to different performance levels (the desktop RTX 4080 uses 320 Watt for comparisons). 

The performance depenends on the TGP settings which defines the maximum boost clock (1455 MHz for 80W and up to 2040 MHz for 150W). With 150W the performance should be in the region of the desktop RTX 4080 and therefore best suited for 4k60 or 2k with 120 fps for demanding games. For raytracing, lower resolutions or DLSS is needed.

In our reviews, an average RTX 4090 is clearly the fastest mobile GPU and bests the RTX 4080 Laptop GPU and mobile RTX 3080 Ti easily. 4090M GPUs with lower power limit can however loose up to 20% performance.

In games, fast RTX 4090 models allow gaming in 4k with maximum details. Adding Raytracing may lead to frame rates below 60 fps (e.g. F1 23) and therefore the use of DLSS or reduced settings are advised.

The AD103 chip of the RTX 4090 mobile is manufactured in an improved 5 nm process at TSMC (4N).

NVIDIA GeForce MX250

The Nvidia GeForce MX250 is a dedicated entry-level mobile graphics card for laptops. It is based on the same Pascal GP108 chip as the predecessor, the GeForce MX150 / desktop GeForce GT 1030 but features increased clock speeds. As with the MX150, the MX250 is available in two versions, the normal 25 Watt version (1D13 device ID) and a low power version with 10 Watt TDP (1D25 device ID) and reduced performance (The MX150 with 10W had a 32% lower clock speed). The GDDR5 memory interface now also supports speeds up to 3.5 GHz = 7 GHz effective (up from 3 GHz).

The GP108 chip is manufactured in a 14 nm FinFET process at Samsung (LPP) and offers a number of new features, including support for DisplayPort 1.4 (ready), HDMI 2.0b, HDR and improved H.265 video de- and encoding (PlayReady 3.0). 4K HDR with Netflix however wont run due to the minimum 3 GB graphics memory that is required. Compared to the bigger Pascal chips, the small GP108 does not support Simultaneous Multi-Projection (SMP) for VR and G-Sync. A full list of improvements and the new Pascal desktop GPUs is available in our dedicated Pascal architecture article.

Thanks to the increased core speeds (up to +3% in Boost for the 25 Watt version, and faster memory speeds), the new MX250 should be slightly faster than the older MX150. First benchmarks indicate a 5% increase. Still modern should only run fluent in low settings (see the page of the MX150 for games and benchmarks).