Start a Conversation

Unsolved

This post is more than 5 years old

F

14118

October 21st, 2017 09:00

Run an application on the second screen using the nvidia video card (d3100 docking station)

Hi, guys,

I have next setup: Dell Inspiron 7567 laptop, Dell D3100 docking station and LG Flatron e2360 monitor. And I have a problem with it.

The problem is related to video cards installed on the laptop. The laptop is equipped with Intel HD Graphics 630 and NVIDIA GeForce GTX 1050 Ti. And the problem is that laptop/docking station/whatever doesn't use NVIDIA card to display the picture on the second monitor. So, if I run game or movie on second monitor it's displayed using only Intel card. Problem doesn't appear if I connect everything directly into the laptop(I guess, it's a docking station problem). How to solve that problem?

12 Posts

October 23rd, 2017 13:00

Here is answer from DispalyLink team. :(
displaylink.org/.../showpost.php

9 Legend

 • 

14K Posts

October 23rd, 2017 14:00

As hinted at by the above post, the D3100 uses a DisplayLink chip, which means displays connected to it are not directly driven by either of the GPUs in your system.  Instead, the CPU and GPU render display frames, then compress the data and transmit it as regular USB data traffic to the dock, where the DisplayLink chip decompresses it.  This can lead to some undesirable compression artifacts, such as blockiness or inconsistent/slow motion under certain conditions, including when large portions of the display area are changing simultaneously (full screen video, gaming, etc.), or when there is bandwidth contention over USB (when transferring files to a hard drive), or when the CPU is already busy with some other tasks.  The extra compression overhead also of course takes a performance toll.

That said, the 7567 does have a built-in HDMI output, which would be directly driven by the GPU and therefore would not introduce any of these problems if you had your display connected that way, although of course it means a separate cable connection.  Unfortunately, that seems to be the ONLY built-in display output available on that system, so running multiple independent external displays would not be possible without a DisplayLink solution like you have now.  Doing that while maintaining native GPU control of the displays would require your system to have a DisplayPort, USB-C (with video support), or Thunderbolt 3 connector.

Also note that the newer D6000 also uses a DisplayLink chip, even when connected via USB-C.  Normally, using USB-C means that the dock/device is tapping into the native GPU DisplayPort output wired into that port, but the D6000 doesn't work that way because it's designed to be compatible with USB-C and USB-A systems.  The only advantage to using USB-C with the D6000 is that it's able to provide up to 60W of power to systems when connected that way.

9 Legend

 • 

14K Posts

October 23rd, 2017 14:00

And to add to the reply you received in the DisplayLink forum, the only Dell laptops I'm aware of that allow you to disable the integrated GPU and use the discrete GPU exclusively are the Precision 7000 series.  In many other systems, the discrete GPU often isn't even directly wired to display outputs.  Instead, it functions as a render-only device and sends completed video frames to the integrated GPU, which then passes them through to the connected display(s).  With this design, it wouldn't even be possible to have the discrete GPU drive the displays since there's no physical connection there to begin with.  The Precision 7000 Series models actually have DisplayPort multiplexers built into them so that both GPUs are wired to the multiplexers, and the multiplexers are wired to the output connectors, and then a BIOS setting determines which GPU's connection is actually allowed through the multiplexer to the outputs at any given time.  That design is obviously more complicated and more expensive, and the scenarios that require giving the discrete GPU direct control of the display outputs are pretty limited, which is why it's not commonly available.

12 Posts

October 23rd, 2017 15:00

"In many other systems, the discrete GPU often isn't even directly wired to display outputs.  Instead, it functions as a render-only device and sends completed video frames to the integrated GPU, which then passes them through to the connected display(s)"

I guess, this makes using of "the Windows 10 Feedback Hub for request that Microsoft does NOT use the integrated GPU but the best GPU for indirect displays" meaningless.

But I'll try anyway. Who knows what will happen.

12 Posts

October 23rd, 2017 16:00

Here is request on the Windows 10 Feedback Hub.
"Windows 10 uses the integrated GPU (instead the best GPU) for indirect displays": 

https://aka.ms/Cojpi9

Need your upvotes. :)

9 Legend

 • 

14K Posts

October 23rd, 2017 16:00

"In many other systems, the discrete GPU often isn't even directly wired to display outputs.  Instead, it functions as a render-only device and sends completed video frames to the integrated GPU, which then passes them through to the connected display(s)"

I guess, this makes using of "the Windows 10 Feedback Hub for request that Microsoft does NOT use the integrated GPU but the best GPU for indirect displays" meaningless.

But I'll try anyway. Who knows what will happen.



That hardware design doesn't make this request meaningless at all.  Remember, DisplayLink displays technically aren't attached to a "real" display output in the first place, so it wouldn't matter at all that the discrete GPU isn't wired to any display outputs; after all, the integrated GPU being used today isn't directly attached to those displays either! :)

But not having a direct connection to display outputs wouldn't matter anyway.  Remember, even today on systems where all outputs are physically connected to the iGPU, the discrete GPU can still accelerate video performance by operating as a render-only device and feeding completed video frames to the iGPU, as I already described.  So reading the DisplayLink response, it sounds like the issue is that currently, displays attached via DisplayLink cannot do that; instead, they only can only use the iGPU to render frames (and then compress that data for transmission over USB), so the change that needs to happen is to allow the discrete GPU to be tapped as a render-only device even for DisplayLink displays, not just displays directly connected to the iGPU.

No Events found!

Top