Black screen while playing EU

What's happening:

Starting with VU16.0 on Feb 19 (can be a coincidence) my PC sometimes randomly goes black screen while playing EU. There are 2 possible states:

1. the game's sound is still active (mobs attacking, people globalling etc); it's possible to connect to the PC via FTP and VNC (but VNC gives only black screen); pressing Win+Ctrl+Shift+B (restart the gfx driver) either makes the black screen flicker slightly without any changes or pushes the machine into state 2.
2. 
   
3. no sound, no external connection possible, apparently it hangs.

It stays in those states indefinitely until I press reset.

The PC runs Entropia for 10-24 hours a day (with overnight crafting) and this can happen once or twice (rarely) a day or not occur for 2-4 days. It never happened so far during overnight crafting sessions. There is no visible pattern to it, can happen in the middle of a hunt or after a prolonged standing on same spot chatting.

It never happens outside of Entropia. Other games I've played in these months are Metro Exodus and Fallout 4.

The machine:

CPU: Intel i7 5820k, cooled with Noctua NH-D15
MB: ASUS X99-DELUXE
RAM: 16Gb DDR4 2666MHz Corsair Vengeance LPX (CMK16GX4M4A2666C16)
GPU: GeForce GTX980 ASUS DirectCU II PCI-E 4096Mb (STRIX-GTX980-DC2OC-4GD5)
SSD: 1Tb Samsung 850 Pro
PSU: Seasonic Platinum SS-1000XP 1000W
Case: Silverstone Fortress FT02B
OS: Windows 10 x64 Pro 1809

Runs 24/7 since 2014 with zero problems (until now). Normally it's overclocked (3.3 > 4.5Ghz) but is reverted to stock since the problem started to occur. Entropia runs at 2560x1440 windowed fullscreen all at max.

What I've done so far:

- reinstalled Windows and EU few times without using any old backups or saved content
- tried running EU on a bare freshly installed Windows without installing anything else
- made sure all the drivers and redistributables are the latest
- tried every Nvidia driver from the past 3 months and also a couple of year+ old ones

- 12 hours of stressing with each of: Prime95, OCCT, memtest86, FurMark, Unigine's Valley/Superposition. No slightest problems.
- there is no overheating, OCCT logs everything continuously and the temps/voltages/etc logs don't show anything out of ordinary prior to an incident
- there are no errors or warnings in the Event Viewer other than the expected "Windows was not properly shut down" on booting back, nothing in the Minidump folder (so apparently no BSOD)

What I can't do:

All machines I have access to are either macs or laptops so I can't test the PC with a different gfx card. And should I even doubt the card anyway after all those stress tests?


In before "omg ditch that antiquated bucket of bolts and build a new one", the machine performs splendidly and still exceeds my needs by far. Can't see me replacing it in the next 2-3 years, especially with the current Intel's difficulties (and I would rather not touch anything of that other brand). Also, don't want to jump to conclusions but by the look of it I'm not entirely convinced that the problem is not on Entropia's side.

It's not a huge deal yet, I just reset and move on (although it often costs me a mob), just afraid that it can get worse.

Thoughts?

EDIT: In the end, it was just slowly dying gtx980. Symptoms eventually worsened but getting a new rtx2080 solved everything.

I'm in a hurry but, is it possible the microcode fixes for spectre and meltdown are screwing with your CPU? You could look into that.

Some people choose to uninstall and block certain KB updates for windows 10 rather than be safe against spectre/meltdown due to problems like this (iirc).

Other than that... dying PSU maybe?

PS: I know this might sound dumb but... did you clean the computer of dust and such?

Angel O2 Mercer said:

Other than that... dying PSU maybe?

Click to expand...

But it passes days of stress tests and it doesn't occur outside of EU.

Angel O2 Mercer said:

PS: I know this might sound dumb but... did you clean the computer of dust and such?

Click to expand...

Yes, it's perfectly clean.

Will look into the spectre/meltdown fixes.

Svarog said:

But it passes days of stress tests and it doesn't occur outside of EU.

Click to expand...

Hmm, right. Didn't read thoroughly.

Other things I would try:

1) Replace the CMOS battery (CR2032). They are like 50 cents where I live, and they have saved my butt more than once.

2) Test the computer with only one RAM stick (test this with all your sticks separately and in different RAM slots). Sometimes only one of your sticks (or RAM slots) is damaged, and it doesn't mean that memtest86 would detect it. (unlikely though)

3) Move the GPU to other PCIE slot (if you have more than one).

If we can rule out any hardware problem the fix should be cheap

Cheers.

"Tech support loves this man, click here to find out why"

I would try the laptop and keep the game running for the same amount of hours and see if you can replicate with another machine

EDIT: Maybe move the game from the SSD to a HDD just to try