5

I'm running a laptop with an nvidia mx150 graphics card on Ubuntu 17.10. I noticed my video was very slow while playing video. So, I started looking at the video driver, and running glmark2 to benchmark the graphics.

I tested with 3 different drivers: nvidia 387.34 (open source) nvidia 384.11 (proprietary) X.org Noveau

For each of these drivers, the performance was similar. I was surprised as, in the past, the nvidia proprietary drivers provided a significant improvement in video performance. Maybe the MX150 graphics card doesn't use features that require the proprietary driver (and may be why this card has 1/4 the performance of the GeForce GTX 950)

Since the drivers didn't seem to be the cause, I decided to switch from the wayland manager to Xorg. This is where the video performance improved significantly. glmark2 with xorg (above drivers): 2375-2460 glmark2 with wayland (above drivers): about 1480

So, wayland cuts video performance by about 40%. This is pretty significant. I'd switch back to Xorg, but the one feature of wayland that is very useful is that I can have fractional scaling for each display. When dealing with a high-res laptop (my HP spectre has a 3840x2160 display) and a bigger secondary monitor, having separate scaling with finer controls than 1, 2, 3, etc., is very useful.

At this point, I think I just need to live with this crummy performance, or switch to Xorg if I need that additional performance and deal the over-sized images on the secondary monitor.

Is this information useful to the Ubuntu or wayland development team? Should I presume they're bench marking wayland with drivers similar to the MX150? If it might be helpful to them, how would I inform the wayland and Ubuntu teams of this data?

0

You must log in to answer this question.

Browse other questions tagged .