Hi,
I've noticed something that makes me think this is not a graphics problem. Just lately, whenever I notice a pause or stutter I've checked out the net status and seem to see that it correlates with a simultaneous spike in the "Variance in delay". A pause corresponds to a large spike and stutters with a series of smaller spikes. Often they come together.
So, it looks like the graphic effects I'm seeing are linked to the variance in delay and not my graphics set up. When it happens it happens suddenly and only lasts a fraction of a second but the net status shows a history over the last several seconds so its easy to check. I'm also trying to stay alert to see if the pauses also correspond to a fluctuation in frame rate or if that stays high through the pauses and stutters. Difficult though because its all over so quickly.
The plot thickens.
Any insights into this greatly appreciated.
BigMol
Hi Mol,
I have provided an attachment below that I had captured using MSI Afterburner's graphing overlay while playing AHIII DX11 vers some time back that shows that the screen freezes do interrupt the GPU directly (looks like the GPU is entering a TDR-timeout delay recovery-GPU stalls then the monitor drivers call the OS to reset the GPU drivers to restart the GPU to resume operation....all this happens within the 2 sec window the OS allows before calling a fault and AHIII generating a .dmp file).
This is related to a driver issue running under Directx 11 API at the OS kernel level and I still get these to date running on the current box in sig below using the latest graphics drivers.
I have also provided some attachments that I also captured using Latency Mon running in the background while playing AHIII DX11 vers that also caught my box going thru a screen freeze. The attachments show that the OS Kernel Mode Driver Framework (driver file Wdf01000.sys) was getting a ton of ISR's which set off a bunch of DPC's which will stop a process from utilizing the CPU until the DPC has finished executing.
The issue is, which hardware driver is the real issue that is causing the OS to halt AHIII momentarily while operating under DX11? Or is the issue within Windows OS itself? From looking at the screen it seems obvious that it appears to be the GPU drivers, but it could also be the sound chip drivers, or LAN drivers (or a combination of all of these) that is being affected but the issue is at the Windows OS kernel level in conjunction w\ operations thru the OS level DX11 API which primarily intercepts the game's generated graphics calls then sends them after applying the called for Directx 3D graphics applications to the GPU drivers to then instruct the GPU to process the calls then store the finished frames in mem buffer until called to flip them to the monitor screen. Thruout all these episodes I've never heard the game's sound stop while the screen is froze or have noted the ingame vox to stall while this is going on but I don't use vox most of the time so I can't personally say but I know several players have reported that vox is interrupted when a screen freeze occurs which lends to what you've noticed.
Under Dx9 this is mostly gone as when I've run the game under Dx9 (the game client's coding is the same being used w\ both Dx9 and Dx11 API's...the 2 game shortcuts when activated instruct the OS as to which Dx API to use w\ the game) this issue is non existent. Now under Dx9 the GPU is put under much more of a graphics load processing the in game graphics calls than it is processing the same graphics calls under Dx11 but under Dx9 these screen freezes are mostly non existent whereas under Dx11 they are occurring at random intervals for .5 secs to 1 sec....I have witnessed them happening when the GPU was under a heavy load and also while relatively unloaded w\ no set pattern established except that each time they do occur the GPU is momentarily stalled as shown on the graph (GPU usage %, GPU framerate and GPU frametime drops towards 0 then restarts and resumes normal operations).
As you can see from the dates on these snippets this issue has been going on w\ AHIII for the past 2+ yrs when using under Dx11. So while HTC and Co is trying to figure this out, most just run the game under Dx9 to not have to deal w\ the issue.
Just trying to give you a little history background concerning this issue playing AHIII, at least from my perspective.
Now if you do figure out something that we haven't found yet please let us know as I would appreciate the efforts.
Hope this helps.
