February 21, 2012

Status of the Apparatus

Skyrim being the appartus.

As of right now, Skyrim seems to be stable on my system.  Actually, not to put too fine a point on it, Skyrim is working with my current save file.  After all the testing I've done in the past several months, I've determined that my hardware is not the issue.  The problem that causes Skyrim to crash my system seems to be related to the game itself.  In interviews published last week, Bethesda's Skyrim director Todd Howard gave some insight to some of the issues that Skyrim developers have been aware of for some time now.  Specifically, it is NOT related to large save files:

"That’s the common misconception. It’s literally the things you’ve done in what order and what’s running. Some of the things are literally what spells do you have hot-keyed? Because, as you switch to them, they handle memory differently."

"We tried doing it through e-mail. We need to open the saved game comes up and look at it. We’ve got one guy who has seven dragons on the other side of the world, and a siege about to happen in this city and another 20 quests running. And, ok, this is what the game is trying to do and it’s having a hard time running that."

Not All Crashes Are Created Equal

After countless forum/website/blog reading, there seems to be several types of crashes.  Listed below are the ones I've categorized in my head as well as the theory I have and the suggested actions.  At the bottom of this post are some links to tools and some information on using them.  Note the information below (except the Skyrim specific ones) can be applied to any game or program that is giving you heat (sorry for the pun).

Crash to Desktop without Error Message - this happens while playing Skyrim and the TESV.EXE application just exists and dumps you to the desktop.
Theory: This is likely due to some kind of corruption with the game. 
What to do:
  • Stop overclocking CPU, RAM and Video Card (just to be able to test with a default setup)
  • Run RAM stress test to look for memory errors.  Replace defective RAM modules if necessary.
  • Run Video Card stress test to look for artifacts, crashing, etc. 
  • Upgrade Video Card drivers to latest STABLE NON-BETA release.
  • Replace defective video card if necessary.
  • Check for overheating

Crash to Desktop with Error Message - the error message is related to the video driver crashing and recovering
Theory: Something caused the video card driver to crash
What to do:
  • Stop overclocking CPU, RAM and Video Card (just to be able to test with a default setup)
  • Run Video Card stress test to test stability of video card and drivers
  • Upgrade Video Card drivers to latest STABLE NON-BETA release.
  • Check for overheating

Crash to Blue Screen - the standard Windows OS crash with a blue screen background with lots of data on the screen.  You all know this one, I'm sure.
Theory: Normally, a driver crashed, but the cause could be a bad driver or hardware the driver was talking to.
What to do:
  • Stop overclocking CPU, RAM and Video Card (just to be able to test with a default setup)
  • Identify the driver by looking at the Bluescreen or the dump files and test the driver and hardware associated with driver.
  • Definitely get help on this one.
  • Also, check to ensure that your Power Supply can keep up with your system.
  • Check for overheating

Crash to Black Screen - this is what I've seen - it's a crash where the monitor loses signal and the sound loops and sometimes reboots your whole system.  Sometimes you have to power off manually.
Theory: Something in the game engine causes the video card hardware and/or RAM to access a part of memory that does not exist (computers don't like that). 
What to do:
  • Stop overclocking CPU, RAM and Video Card (just to be able to test with a default setup)
  • First, rule out the hardware:
  • Run RAM Stress Test
  • Run Video Stress Test
  • Run CPU Stress Test
  • Run the above overnight
  • Check that your power supply can keep up with your system
  • Check for overheating
  • Fix any issues identified
If all the above are good, then do the following:
  • Turn off the system and let it sit for a minute.  Also, to be sure, try removing the power cable for a few minutes to ensure that any artifacts are removed from memory.
  • Power on the system.
  • Start Skyrim.
  • Turn off ALL Auto Saving.
  • Reload your last saved game.
  • Clean House!  This means remove any favorites that you don't use regularly, unbind unneeded hotkeys, get rid of ingredients and potions and anything else you are carrying around that you don't need (put them in a chest in your house or just drop them somewhere).
  • Create a NEW save game.
  • Exit Skyrim.
  • Start Skyrim.
  • Load that new save game.
Generally, if you suspect it's the last item above (i.e. a save file that is causing the crash), you'll want to keep that save file (make a copy) and then use that repeatedly.  For example, load it, make it crash so you know exactly what to do to make it crash and then restart, load it, make a change and see if it crashes at the same place.  If you get beyond it then do it again, don't do what you just did and see if it still crashes.  Then make the same change and see if it doesn't crash again.  In other words, make sure it's repeatable and demonstrable so you can pinpoint the change that makes it not crash.

Video card drivers - just go to your manufacturer's website and find the appropriate driver.  Be sure to do a "clean install" if possible - Nvidia driver installation has this option during the advanced install.

Hardware Testers:
Prime95 - Let this run at the highest settings overnight.  Be sure to test all of your RAM.  Any RAM issues should surface.
Linx - Another RAM tester.  Same as above - let it run against all your RAM overnight.
OCCT - http://www.ocbase.com/perestroika_en/index.php?Download - this tests multiple aspects of your system. Use it to test your GPU RAM and your CPU.  Again, let it run for a long time to encourage any hardware issues to surface.
Kombustor - this will test your GPU in DX9, DX10 and DX11 modes (test each one to be sure) - let this run overnight and any errors should surface.
Heaven - this one tests your game rendering.  Let this run overnight.  You can also choose DX9, DX10 and DX11 tests.  I suggest running them all, one after the other.

If you do the above tests, AT THE VERY LEAST you will feel good about your hardware being solid.  I've run the first four items AT THE SAME TIME for hours and hours and no issues surfaced for me and yet I experienced Black Screen crashes.

Monitoring for Overheating
I strongly recommend using HWiNFO (32 or 64 depending on your OS).  Start this and set it to log your stats to a file.   If you do crash, you can go back and check many, many things, like your CPU utilization, CPU temperature, System Temperature, GPU Temperature, GPU utilization, RAM usage, GPU RAM usage, etc, etc.  It's a very handy tool.  Feel free to send me your log files if you need help figuring out what it means.

Remember, this is an awesome game and there is a lot of mis-information out there.  When troubleshooting, you always want to get to the root cause of the issue.  Software developers will always blame your hardware first and hardware manufacturers will always blame the software.  Forget them - get to the truth yourself, learn something in the process and hopefully at the end you can enjoy your game.

No comments:

Post a Comment

Please be helpful and courteous. Thanks!