The story behind this:
My old graphic card recently stopped working and I was forced to buy a new one (GTX 760), since then my PC started crashing while playing GW2. I’ve already seen some threads considering the same problem with different Nvidia cards, but I haven’t seen a working solution so far, so I hope to specify this error and it’s source better.
Also one thing to mention is that I only get those crashes in PvP, neither in PvE nor in WvW (most likely because I reduce the gfx-settings for WvW).
(I additionally use the SweetFx shader injector, but this is NOT the cause of the crash, since it makes absolutely no difference in the crashing intervals etc. I also disabled this for most of the testing)
The crash details:
The screen turns into a single color (I already had black, white purple, blue and pink…), the PC stops responding (hardcrash) and the soundcard plays the last miliseconds in a loop.
I also had some softcrashes where the screen also turned into a color, but the PC started responding again after a few seconds, the sound usually continues normal except a short stuttering (through one softcrash i even managed to get a GW crash-report, yay).
So where does the error come from?
It is most unlikely caused by the Rig (gfx-card/PSU) as some people have suggestet all over the forum, and I’ve found a few indicators for that. I personally think it is either a bug in the game or in the driver.
Testing 1: Why it is most unlikely to be the graphic card:
First of all, I’ve run Battlefield 3 and the Battlefield 4 beta on high/ultra specs (msaa turned off, of course) and I had no crashes there, while my Rig gets hotter than in GW.
Second, I ran some tests (GPU-Z logs attached):
- chkdsk and MemTest to ensure that the error is not caused somewhere other than expected
- FurMark to ensure the gfx-card can stand the temperature and the stress
- memtestG80 to ensure that the gfx-card memory is fine
- OOCT as another stress thest.
The results are:
- there have been no crashes while benchmarking/testing
- the HDD, RAM, gfx-card-RAM are fine
- the gfx-card can handle the benchmarks without problems
Testing 2: Why it is most unlikely to be the PSU:
Not knowing how to test the PSU better than to max out the Settings and additionally use SweetFx (also tried this without SweetFx, it makes no difference), I did this and played a few rounds of PvP. This should normally increase the power-consumption of the whole PC and if the PSU was bad it would definitely crash the System, at least in my opinion. Expecting the PC to crash i played for about 2 hours PvP, after that a bit WvW (on max settings) and some PvE (Mad Labyrinth zerging) and I didn’t get a single crash. The range of FPS was from 6 to about 50 (50 only in almost empty areas). At the moment I’m not quite sure if the game does not crash on these settings at all or just later (normally I have these crashes after about 2 rounds of PvP), but it clearly can’t be missing power of my PSU.
Logging the stats:
As I wrote I logged the benchmarks/teste with GPU-Z, but I also managed to log a hardcrash, as well as a softcrash and the normal GW running without a crash (I can’t detect a significant cause for the crash there, but one can see the softcrash beeing logged in that file)
The log-files are:
- GPU-Z Sensor Log_furmark.txt the FurMark log
- GPU-Z Sensor Log_memtestG80.txt the GPU-memtest log
- GPU-Z Sensor Log_ooct.txt the OOCT log
- GPU-Z Sensor Log_gw2_(crash) crash at my normal settings
- GPU-Z Sensor Log_gw2_(nocrash) no crash at my normal settings
- GPU-Z Sensor Log_gw2_ss_sfx_high_(semi-crash) softcrash with SweetFx, supersampling and high (not max) settings
- softcrash_report.txt the GW crash-report I managed to recieve at a softcrash
(note: these log files are a bit shortened)
The windows event-viewer only shows “error 41, Kernel power” and nothing more.
