eGPU works fine using built-in display, but not using HDMI
 
Notifications
Clear all

eGPU works fine using built-in display, but not using HDMI  

  RSS

(@guilherme_vilhena)
Active Member
Joined: 4 months ago
 

Hello everyone,

System specs

  • MSI GE62 6QD 15.6-inch
  • CPU: Intel Core i7-6700HQ
  • iGPU: Intel 530
  • dGPU: GTX 960M
  • OS: Windows 10

eGPU hardware 

I have installed the eGPU to my machine and installed the most recent drives, but I am facing a funny problem:

When the eGPU is under load (game or benchmark) and with the HDMI cable connect directly from it to the monitor it starts to freeze. On games, for example, the screen starts to blink and of a sudden go black (it blink sometimes with the game image, and black for long periods again, and stay black). I have realized that when changed graphics setting from Highest to Lowest this problem is not so often. By the way, the game I was testing was Shadow of the Tomb Raider.

Another funny fact is that when I disconnect the HDMI from the eGPU and start to run the game on the built-in display, this problem does exist. If I connect the HDMI to the motherboard HDMI output, the problem returns.

I have disabled the  NVIDIA driver , but it does not solve it. I also disable the Intel 530 graphics, but no success at all.

The problem is not the HDMI cable, since I have used it until now, and it works perfectly.

I can see in  GPU-z  that the port is working as it should (PCIe 16x 3.0)

Does anyone have any clue?

Thank you in advance!

Edit -------------------------------------------

I have played with some configs from Radeon Software and realize that some of the changes make the problem be less frequent. For example, I changed the options for the

- Radeon Enhanced Sync

-Surface Format Optimization

-Virtual Super Resolution

-GPU scaling

-HDMI link assurance

The best composition with these options that I found was with the Radeon Enhanced Sync and the Surface Format Optimization both turned on, meanwhile, others were off. This config make the problem less frequent, with a mean of appearance around every 75 seconds (yes, I calculated it with a chronometer). This result was far better than any other setting, which I have played a lot. 

With other combinations, the frequency of the problem had a mean of 20-40 seconds.

I did also a GPU stress, and a test to see power consumption. GPU is working fine, as well as power consumption. 

Meanwhile, I could realize that HWinfo reported errors for the PCI/PCIe bus, and the number of errors was increasing every time the problem comes up. Every time the screen blinked or goes totally dark, I did an alt + tab and checked the number of errors. They pretty much matched. It is like 2 blinks corresponding to +2 errors reported.

Now I am on the step to find these reported errors on windows (first step to get things neat). If you have any idea, just let me know, every help is welcome.

Edit 2 ----------------------------------

Finally, I have found what probably is causing these errors by the Event Viewer. The complete report of the error is:

"

A corrected hardware error has occurred.

Component: PCI Express Root Port
Error Source: Advanced Error Reporting (PCI Express)

Primary Bus:Device:Function: 0x0:0x1D:0x0
Secondary Bus:Device:Function: 0x0:0x0:0x0
Primary Device Name:PCI\VEN_8086&DEV_A118&SUBSYS_115A1462&REV_F1
Secondary Device Name:

"

I have checked Google on how to prevent this error from coming up again, but I was not successful. Could someone give a little help?

 

Here is the performance of the eGPU, everything to be very normal...

ainda 64
This topic was modified 4 months ago

Pending: Add my system information and expected eGPU configuration to my signature to give context to my posts


ReplyQuote