I am having trouble with the graphics device disconnecting with the reason DXGI_ERROR_DRIVER_INTERNAL_ERROR on some of our devices after our game terminal has been running for a few hours. Specifically the Intel HD 530, though a few of the other Intel based GPUs will crash as well. The odd thing is that the same code will run indefinitely on the same computers if we put in a discreet card. It also runs without any problem on other Intel chipsets, even those that are quite a bit older. We have tried the basics like Windows updates and getting the latest drivers from Intel. We have also verified that this is not a computer-specific problem, as the problem persists across different computers with the same build.
I have tried adding ID3D11DeviceContext::Flush calls by resource creation as suggested by https://software.intel.com/en-us/forums/graphics-driver-bug-reporting/topic/610376 with no apparent help. I have also verified that no graphics handles are being held onto for a very long time, and our GPU memory usage never gets much above 400 Mb, which should be well within the ability of an integrated card to handle. We actually wrote a watchdog application to monitor that, and the usually the device is removed while the memory is lower than normal.
I'm having a hard time finding any resources that would help us find the root problem, as DXGI_ERROR_DRIVER_INTERNAL_ERROR is not a very helpful error message.
We are using the DirectX 11 api, and run on a variety of Windows based computers, including both all in ones and desktops.
I would appreciate any help or ideas anyone has, as we haven't been able to make much forward progress even after a few weeks of intensive debugging and engine changes.