I have been experimenting with UE4 since the free release, and have had no problems. However, a few days ago I had a BSOD (not related to UE4) which I traced to driver corruption.
I updated my video card (a Mobility Radeon HD 5650 1GB) to the latest stable release (14.12), and since then I cannot open any UE project. It gives me the error “The Video Driver Crashed and Was Reset”.
I am not experiencing trouble with any other graphics programs or games, including UE3-based games such as “Life is Strange”. They do not seem to exhibit any artifacts, and Windows does not seem to think anything is wrong. Temperatures are fairly typical for this laptop, although it is the summer to obviously cooling is a little more problematic than normal (particularly since Acer seems to think laptop cooling is unnecessary). Still, I tried loading UE4 editor while monitoring temps using SpeedFan and I did not see any reason for the Video Card driver to crash - it was at 56C while Discovering Assets.
However, I did notice something odd after the crash. The launcher itself seemed to suddenly have artifacts (not the rest of the system, however). I am attaching a screencap of the relevant window:
I am also attaching my dxdiag: [link text][2]
Note that the screencap has not been altered in any way except to crop to fit only the relevant window.
If it weren’t for the fact that the Editor had worked fine before, I would think that my computer is too weak. But I was able to do the 3rd person tutorial prior to this, and now even that isn’t loading. Is there a known conflict between Catalyst 14.12 and UE 4.7.6? Or is there something else going on?
Hey Drafonis,
I’ve had a few of our EngineQA guys try out the 14.12 drivers and tried it myself on one of our AMD GPU laptops and a Compat Desktop. Unfortunately, we don’t have any AMD GPU laptops with Win8.1, but the Compat Desktop had 8.1. I was unable to reproduce this issue on any of these machines.
Does the engine ever throw a callstack? This may be an issue worth reporting to AMD as it may be specific to your laptop GPU.
-.
,
I think I found what’s going on by examining the last log from my current project. Here are the relevant log segments:
[2015.05.27-03.51.51:467][133]LogD3D11RHI: Timed out while waiting for GPU to catch up. (0.5 s)
[2015.05.27-03.51.51:969][134]LogD3D11RHI: Timed out while waiting for GPU to catch up. (0.5 s)
[2015.05.27-03.51.53:575][135]LogD3D11RHI:Error: Direct3DDevice->CreateTexture2D(TextureDesc,SubResourceData,OutTexture2D) failed
at D:\BuildFarm\buildmachine_++depot+UE4-Releases+4.7\Engine\Source\Runtime\Windows\D3D11RHI\Private\D3D11Texture.cpp:433
with error DXGI_ERROR_DEVICE_REMOVED
Afterwards, the editor called FPlatformMisc::RequestExit(1) and the log closed. It seems odd that it talks about D:\BuildFarm, as such a directory does not even exist on my computer (unless I’m misunderstanding something). The project itself is located in my Documents library, and Unreal is in Program Files. Could this be significant, or am I barking up the wrong tree? I am doing a fully local build using Visual Studio 2013 Pro SP3.
D:\BuildFarm is where D3D11Texture.cpp was last compiled, which is on our BuildFarm.
Anyways, would you mind sending me the entire log txt file? I’ll forward it over to one of our engineers to take a look.
-.
Marking this as answered due to inactivity. Please respond to this answer to re-open.
-.
,
I apologize for the delay in replying. I had a family situation which prevented me from responding in a timely manner. I am attaching the log as requested.
Hey Drafonis,
I talked with one of our engineers and he suspects it could be the GPU going bad. However, I’d like to see if you could roll back your drivers to a previous version and see if it works. If it doesn’t improve, it could be worth reaching out to AMD.
-.
,
I haven’t rolled back my drivers, but, on a whim that came to me before I went to bed last night, I tried reloading the project since I had switched my laptop cooler to a better one than the one I had been using earlier (the ball bearings had been going bad, causing the idle temps to be about 20 degrees higher than they are now).
Lo and behold, the project loaded and I was able to to make changes to the project (such as editing textures). As such, it was likely that the crash was occurring due to the GPU throttling down due to mild overheating, which I should have suspected due to the error from the log (“Timed out”).
Regardless, the error was clearly on my end rather than Unreal’s or AMD’s.
Thank you for your help and that of the engineering staff. I definitely owe you all one for your dedication, as I was going stark mad trying to solve this.
