I've been randomly having Folding units cause blue screen restarts off and on since I started Folding back in March.  After finding out early that others had had problems with bad work units I always just wrote it off as that.  Today I actually caught one of the crashes (usually it's overnight) and notice the displays just went all corrupted with alternating light/dark rectangles and graphical anomalies covering the desktops. 
 
It was suggest to run the OCCT GPU tests so I did.  3D test ran for 45 minutes no issues.  VRAM test causes the issue in seconds.  I tried underclocking the VRAM by 500 and it lasted nearly 6 minutes and 12 passes then it happened again. 
 
I'm not sure if that's just an OCCT issue, driver issue (run 456.38) or even potentially an issue cause by running a PCIE extender due to vertical mounting.  The extender is the Phanteks 220mm with the extra shielding and the new design where it's split in a number of individual sections.  I can't test without since I'm watercooled.  At least not without draining the loop, dissembling the GPU section, and finding some extra tubing to reconfigure it.
 
I've had zero issues in real content like gaming, benchmarks like Firestrike, etc.  Just trying to figure out if I have a real issue, or the FTW3 cards are just pushed to the limit and high stress like Folding or OCCT is able to push it past that limit.  Anyone have any thoughts?
 
edit:  just to add, even with the OCCT crashes, it's not something like the card just went today or anything.  After the OCCT issue it actually ran a large Folding unit through to completion no problems.  So it seems to be just extreme high stress that causes it.

Riser could be the issue. I’d suggest testing without one.

I also posted the following in one of your threads... https://forums.evga.com/FindPost/3060562

Not sure if this helps.  I tried the latest driver, thinking maybe it was a driver issue as it seemed to start happening more frequently again since the one from mid-September got installed.  Didn't fix the OCCT problem so far, but I managed to get some event log errors from it this time.  It's stuff like this, which from what I could find potentially also points to driver issues as much as hardware issues.  For software issues it was pointed out that it occurs under very high stress loads which is my case.  Again, zero issues with games, benchmarks, etc.  Wondering if I should just contact EVGA about RMA anyway but if I do what the heck am I even going to end up with as a replacement since all the 2080's are out of stock and same with the 3xxx cards.
 
These are all nvlddmkm errors for Source, EventID 13
first error is this with eventID 14 though: 
\Device\Video3
20df(3268) 00000000 00000000
 
EventID13:
\Device\Video3
Graphics Exception: ESR 0x52df30=0xb030020 0x52df34=0x4 0x52df28=0x4c1eb72 0x52df2c=0x174
 
\Device\Video3
Graphics SM Global Exception on (GPC 5, TPC 3, SM 0): Multiple Warp Errors
 
\Device\Video3
Graphics SM Global Exception on (GPC 5, TPC 2, SM 0): Multiple Warp Errors
 
\Device\Video3
Graphics SM Warp Exception on (GPC 5, TPC 2, SM 0): MMU NACK Errors
 

Here's what I see when the issue happens.  Anyway, sending support a message and we'll go from there.  Just did 2 hours of WoW with streaming with NVenc2 and no issues again still.  Also, I'm still wondering if it's somehow a driver issue or as some forum posts I've read in places have pointed to a possible AMD AGESA issue and something with the PCIE bus not responding properly.  When I induced it to get the corruption pic, windows actually recovered quickly twice which is an improvement since putting the latest (non-hotfix) driver on early this evening.   You can see in the OCCT window where DWM recovered but lost all the taskbar icons.
 

 

 
 

Sajin
If it was a driver issue you could make the issue go away by installing a really old driver. If it's an amd agesa issue you could fix it by updating/downgrading the version. You could also test the card in another computer that isn't an amd system. I'm guessing underclocking both the core & memory to max negative clocks inside msi afterburner did nothing?

I didn't try the core.  However yesterday I did underclock the VRAM by 500 in Precision and the OCCT VRAM test ran for like 6 minutes before it bombed out.  Not possible to test the card elsewhere even if I had another PC.  It's a Hydro Copper SKU so I don't have an air cooler to put on it and of course it's installed into a cooling loop right now.  I'll try underclocking both right now in Precision and see what happens.  If it fixes it what does that indicate?  My guess though is it probably actually is slowly failing since it seems to be having the issue more often with folding in the last couple months (though I've only been folding since late March).  

Sajin
Could indicate two possible issues. A faulty card, or a possible psu issue. I'm guessing your machine passes the occt power supply test though?

I’ll double check that as well. I’d expect it to. It’s only a year old and I haven’t seen anything odd with the voltages. Also currently max underclock I’m 21 minutes into the VRAM test with no issues.

Sajin
21 minutes in. 

Yep, it ran for 30 minutes 51 iterations before I gave up and stopped it.  No errors or any issues.  Set the clocks back to stock and ran the PSU test for 5 minutes (wasn't sure how long to run it).  CPU topped out at 79C (I really need to get and try the offset brackets for my block for Zen 2 chips) and the GPU topped at 53C which is about the highest I've ever seen it.  Max memory temps was the bank between the GPU chip and the power stages at 53C also.  No errors in the PSU test.  Should I have run that for longer?  If not, any thoughts on what this indicates?  Sounds like you were thinking that meant it's the card from what I understood from your other posts.

Sajin
Psu test should run for at least 30 minutes imo. Still looking to be a card issue atm.

Yep I think you're right.  Just let it run for 30 minutes on the PSU test and no issues.  I also double check  my EAR details so it does look like I have access to cross shipping along with Saturday delivery if needed (obviously not this coming one, just that in general I have access to it).  So if it does need to get replaced at least it's "functional" until then. For now.  Definitely no more Folding at least until I hear back from Support and we figure out what's going to happen.  At least I have a valid warranty, just sucks to have a card fail like this especially in the middle of a new GPU series launch.

Finally heard back on my Support request.  They want me to try modifying this registry entry.  I don't really see how that's going to fix it doing that display corruption, but I guess I'll try it.  Not sure what values to try in there though.  
 
https://docs.microsoft.com/en-us/windows-hardware/drivers/display/tdr-registry-keys#tdrdelay
 

Sajin
Seriously? You shouldn’t have to do that.

Yeah but I'll try it anyway I guess to push it along, no idea what values to try though since all they said was to try changing from the default of 2 and raise it.

Sajin
Just change it from 2 to 8.

I tried 3 and 8.  No change really.  Randomly turned Hardware Accelerated GPU Scheduling back off for a second attempt with 3 (and later 8) and the only difference was somehow it ran for a 2 minute test on 3, but only one time.  The rest all still crashed within 20-30 seconds or less.  I didn't expect HAGS to fix it anyway since I only turned that on since late August as a possible tip for performance in the new MS Flightsim but had had these errors folding well before that.  Just never caught one live until recently to actually understand what was going on.  That screen corruption really put me on edge since I'd expect a Windows/driver error to just be a standard bluescreen.  The two bugcheck dumps I got this evening, obviously a "blue screen" event were captured while there was no display though so yeah. 
 
Anyway, I sent them a reply so I guess we'll see what happens next.

thanks for the update
 
every Folding WU is a different load ... max OC is usually no longer possible ... some WU are power hungry, watch your temps
 
New WU launched last month that will push your GPU ... some posts in Folding section of Forums

Was able to run the vram test for 5 minutes on my card. No issues.

Ran it for 18 minutes. No issues.