AWK16

vgerik1234
Day 1 owner of 3090 FTW3 and it just died today.

It’s sad, really sad and I also think this will continue, whether the card is owned from day 1 or recent, with black lip, there is whey more too all this dying fiasco then meets the eye.
Aside from what I said earlier today that one of my friends card that was over a month old died yesterday, friend that works with me got new card with black lip a week ago! and it died this morning.
 
In my view, for the past 2-3 gen EVGA cards had more issues then any AIB partner.  Remember the burning fiasco and thermal pads quick fix, EVGA cards are not like they used to be, at this point their cards longevity and worthy of investment is undoubtedly questionable, heck I buy card to last for a while especially for that much $
 
Its funny how on jaztwocents video, EVGA FTW card also died/degraded enough that cannot be used and he said the card wasn’t even used that card much:   https://youtu.be/3puInnaG5ok?t=739
 
 

another-user
my 3090 ftw hybrid had about 48hrs of runtime on it when it died.
 
from the moment i installed it and put a load on it, i was getting black screens.   setting a -100mhz clock got it much more stable, but it would still happen.
 
i spoke with evga and we both agreed it may be the psu, so i rmad the psu.  i installed the new psu and reinstalled the 3090 and it ran great for that night.  i benched the hell out of it, ran cyberpunk at max settings for an hour or so... all good.
 
the next day i ran the witcher 3 and it black screened in about 5 mins.  i rebooted and ran witcher 3 again, this time with monitoring logs so i could start really looking into the issue.  within 90 seconds of game play the system shut down with an audible pop.   when i rebooted, i could see the card arc for half a second and then the red led came on at the far right power plug.  i swapped cables around just to confirm and the led stayed at the same location.
 
the psu was a new seasonic focus gold, 850w, and i was using 3 separate cables. 
 
evga instantly approved my rma and sent me a call tag like 20 mins later.
 
i ended up emailing them again later that night and just asked for a refund instead, and they agreed.    i dont want to play this lotto and potential lose a psu or the rest of my system in the process.

Gogod2020
I will make an attempt to address possible problems that could lead to black screen 3080 with the fans revving high so that we can possibly discover the solution:

From my extensive testing in real conditions:
1. The abnormal revving was directly connected to specific actions:
A. Minimizing a window esp a game
B. Loading a game or loading screens
C. Graphic intense areas (spells/many players)
D. Random during idle (rare)
E. Random during gaming (frequent per title)

2. The abnormal revving was sometimes exceeding the 100% of fans. When this happened it was followed by loss of display (DisplayPort 1.4)

3. The abnormal revving ignored the custom fan curves

4. The abnormal revving never happened during benchmarks.

5. The GPU was stable during furmark/prime95/superposition/3dmark

6. The black screens took place mainly during loading screens

7. The GPU temperature was never over 70c

8. The GPU sensors did NOT have more than max 1c difference most of them being same temperature.

9. On the final stage (before I RMA) graphics showed deterioration with big fps loss (-50) on games that were 165 frames and then after black screen crash frames dropped to 7-17.

With all the behavior described let’s see what DIDNT work and what partially worked.

1. Disable pcie link power saving: no
2. Set power to max performance: sometimes
3. Remove sleep from pc: no
4. Set fan curve: no
5. Set hysteresis: no
6. Use MSI AB: no
7. Default fan profile: no
8. Switch to OC bios: no
9. DDU drivers: no
10. Update bios: no
11. Disable gsync: no
12. Control Physx from CPU: no
13. Frame cap games: sometimes
14. Lower graphics: sometimes
15. Under volt the GPU: no
16. Change pcie cables: no
17. Reseat the GPU: no
18. Leave side panel open: no
19. Work the fans over 50%: sometimes
20. Disable hardware acceleration: no
21. Reinstall games: n
22. Reinstall Windows: no
23. Update drivers: no
24. Multiple versions of PX1: no
25. Remove Corsair iCue: sometimes
26. Run sfc /scannow: no
27. Switching DP for HDMI: sometimes
28. Change one DP cable for another: no
29. Windows update: no
30. Lowering DX12 to 11: no
31. Lowering resolution: sometimes
32. Playing very old less demanding games: yes
33. Performing less demanding tasks: yes
34. Idling: yes
35. Lowering native 1440p to 1080p: sometimes
36. Disabling ray tracing: no
37. Restore pc health command: no


The solutions that worked “sometimes” either fixed the problem for a few hours or until next reboot of the pc. The things that “worked” demanded that basically the GPU remain idle or next to idle.

GAMES/PROGRAMS TESTED

WoW (Shadowlands): Many issues, black screen and loss of display. Uncontrollable fan revving at certain spells certain areas certain portals. Lowering specific graphics limited the problem but it came back menacing and crashed the display.
WoW (old maps): uncontrollable fan revving
Division 2: Uncontrollable fan revving at load and when loading maps changing areas
Cyberpunk 2077: Uncontrollable fan revving, black screen and loss of display even 5 min after playing, game crashing at start when settings high, game crashing when minimizing.
GTAV: Some uncontrollable fan revving with settings at full ultra everything.
Forza Horizon 4: Some uncontrollable fan revving, black screen and loss of display at loading screens usually after 1 hour gaming.
Older games (Cod/black ops 1) some uncontrollable fan revving.
Mozilla Firefox browsing: black screen loss of display and DELL logo multiple times when disabled hw acceleration
Chrome browsing. Same as Mozilla.
Idle temperatures: 25c (ambient 22-23)
Working temp: 65-68c (ambient 28c/heating on)

All problems occurred between 55c and 68c. The pattern was: GPU power at 50% (utilization 98-99%) power dropped to 30% because loading screen, upon loading (from nvme mind you so 1-2 second load) fans started screaming (temperature at those 1-2 seconds fluctuated by 5-10 degrees which should be covered by the 15 second hysteresis and the 60 second delay of the custom curve). Fans were custom curve controlled to work at 45%.

The list is not extensive I have done other things also and I might have established a pattern: The GPU seems to “panic” that it may get abnormal temperatures by actions taken and as such revs up the fans to prevent an overheating. Senses that there is no danger and goes back down. However if this happens too fast or too many times in a row it will cause a display collapse.

The GPU does not follow the plan of set curves neither does it respect the limits or the hysteresis or the delay set.

Default is horrible it just revs the fans up to 100% even at 60c.

Installing my old GTX 1060 to the pc I was able to run world of Warcraft at the exactly the same settings (bar ray tracing) with zero noise at exactly the same areas/scenarios that made the 3080 go berserk.


My system:
Intel 10700
Corsair HX1200
GSKILL 32gb cl14
Aorus Elite AC v1.0 (F4 bios, latest)
Windows 10 Pro 64bit (fully updated)
NVIDIA Driver (latest)
PX1: 1.1.1.1
SATA SSD 860 Evo 1TB (carries OS)
NVME 970 Evo plus 1TB (carries the games)
Case BE QUIET 500 DX (2 in 2 out fan set)
Monitor: DELL S2721DGF (1440p)

Andrew_WOT
Considering that both China and Taiwan cards are failing, most likely some design flaw, not manufacturing issue.
How ASUS looks like in comparison, does anyone have any data on their failure rate?

AWK16
My apology, I should have been more careful to read your reason for the listing you posted.

Gogod2020

AWK16
My apology, I should have been more careful to read your reason for the listing you posted.

Don't worry mate all I want is someone to point to a solution. If God forbid my next card suffers the same way its impossible for me to RMA again at such high costs. We should NOT have to pay so much for something that is not our fault.