NVIDIA GeForce MX450 found to be 33.5% faster than the MX350 in gaming — half the memory bandwidth and ROPs of the GTX 1650 but with higher power draw than the GTX 1650 Max-Q

[Redaktion]

NVIDIA recently announced the GeForce MX450 based on the TU117 GTX 1650 GPU but was quiet about the specs or performance. Chinese publication Zhuanlan has managed to take the MX450 for a test ride and found that it offers a very good performance bump over the MX350. However, this is at the expense of reduced memory bandwidth and raster operator (ROPs) count along with surprisingly higher power draws compared to the GTX 1650 Max-Q.

https://www.notebookcheck.net/NVIDIA-GeForce-MX450-found-to-be-33-5-faster-than-the-MX350-in-gaming-half-the-memory-bandwidth-and-ROPs-of-the-GTX-1650-but-with-higher-power-draw-than-the-GTX-1650-Max-Q.489569.0.html

[S.Yu]

That doesn't seem at all worth the extra power draw compared to TGL. One can finally buy the best integrated and pass on the worst discrete knowing it's almost certainly the right tradeoff.

[john11092020]

Nvidia is totally losing the low end market. With nothing else to do it throws all the problematic 1650 chips that probably has, chips that probably need extra voltage to work stable. That's why the extra power consumption. I guess it will be giving those at cost if the OEM also buys some extra hi end GPUs.

[Spunjji]

Why wouldn't they just lower the price on the 1650 instead of gimping it like this? It totally breaks the previously quite respectable efficiency of the design, and these are supposed to be used in low-power notebooks.

At this point Nvidia are just throwing their dirty underwear at the wall to see whether it sticks, and if it doesn't, it goes right back on again.

[A]

Pitiful, the gap between low end gpus and igpus closes yet again.

[8&8]

awesome article, good VGA, maybe we will see next year 96 bit bus with MX550 and better GDDR6 but is very funny see a battle between iGPU and dGPU nowadays.

In 2022 maybe dGPU will disappear with Rembrandt APU (with navi) and enhanced Xe 7nm or with X3D adoption arch.