Thanks! Although I interpret the results a bit differently, I've been eagerly awaiting these benchmarks ever since I learned of Thunderbolt 3 eGPUs and heard rumors of low-power quad-code ultrabooks. As someone who games while travelling for work, I need long battery life with the option to plug for my favorite games. I imagine many college students would have similar uses. I'll probably opt for another brand's 2-in-1 myself but these benchmarks help a lot.
Quote from: hfm on December 27, 2017, 01:04:58
I'm sure the CPU also had something to do with the desktop PC's scores as well as available bandwidth.
And yeah, I would agree that for certain games (Battlefield 1, NFS), the low power CPU becomes the clear bottleneck in these charts. That's the main difference the desktop results differ so much, not its 16 lanes of PCIe bandwidth. Just because a desktop motherboard provides 16 lanes, doesn't mean the attached GPU actually saturates much of that bandwidth.
https://www.gamersnexus.net/guides/2488-pci-e-3-x8-vs-x16-performance-impact-on-gpusEven with just 2 lanes (or its equivalent), performance is only slightly affected. There's been so much outcry about laptops that wire only 2 lanes rather than 4, with most suggesting this results in half the performance. That's inaccurate. But believe whatever you want ...or use a piece of scotch tape (or BIOS controls) to test for yourself as shown here.
https://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_1080_PCI_Express_Scaling/