EVGA

3090 FTW Ultra - six months of driver resets

Author
mthompson2336
New Member
  • Total Posts : 10
  • Reward points : 0
  • Joined: 2020/03/16 01:56:41
  • Status: offline
  • Ribbons : 0
2021/06/18 23:23:33 (permalink)
I snagged a 3090 back in November. I'm looking through my logs and notice that I am seeing steadily more crashes and the only thing they have in common seems to be the card. Sea of Thieves throws "driver reset" errors within minutes of starting. Ark and others frequently see "lost 3D3 device". My event log is full of kernel 117 and 141 errors from nvlddmkm.
 
I've attempted every fix suggestion I can find and I'm getting exhausted. Today I sat down to play a game and realized that there's no longer any game on my steam account that I can reliably run for more than an hour.
 
Does this sound like a bad hardware issue?
 
  • My card isn't overheating. I'm not overclocking, and have experimented with underclocking. My memory isn't even running an xmp profile.
  • I've got tons of power - ax1600i. Lots of cooling - 10 case fans, conservative curves, great airflow.
  • I've tried different resolutions, different bit depths, sync/no sync, tried limiting fps, tried every combination of driver and game settings I can think of.
  • I've reduced peripherals down to only a mouse and keyboard directly into the motherboard.
  • I'm running no monitoring software. I've systematically stripped out ever program and service I can to try and avoid conflicts, including Asus and Corsair junk.
Edit:
Win 10 Edu, 64-bit, UEFI (19043)
GeForce RTX 3090 FTW3 ULTRA (VBIOS 94.02.42.80.AD)
NVIDIA Game Ready Driver (466.77)
Samsung Odyssey G9 (5120 x 1440, 240Hz)
ASUS PRIME X570-PRO (3402)
AMD Ryzen 9 5950X
Corsair Vengeance RGB Pro 64GB (4x16)
SSD #1 Corsair Force MP600 NVMe 1TB
SSD #2 WD_BLACK SN850 NVMe 2TB
Win 10 Build 19043
Corsair AX1600i PSU
 
I update regularly, and this has been happening for a long time, so versions given are just the most recent I've tried.
post edited by mthompson2336 - 2021/06/19 21:09:28
#1

15 Replies Related Threads

    arestavo
    CLASSIFIED ULTRA Member
    • Total Posts : 6916
    • Reward points : 0
    • Joined: 2008/02/06 06:58:57
    • Location: Through the Scary Door
    • Status: offline
    • Ribbons : 76
    Re: 3090 FTW Ultra - six months of driver resets 2021/06/19 04:22:32 (permalink)
    Computer specs help (I'm on mobile, so if you have a signature below the post I dont see it).
     
    With that said, many folks have had crashing issues because their motherboard needs to be updated. Check to see if there's a BIOS update for yours.
    #2
    Sajin
    EVGA Forum Moderator
    • Total Posts : 49168
    • Reward points : 0
    • Joined: 2010/06/07 21:11:51
    • Location: Texas, USA.
    • Status: offline
    • Ribbons : 199
    Re: 3090 FTW Ultra - six months of driver resets 2021/06/19 08:27:33 (permalink)
    Yes, it does sound like a bad card. Does underclocking the core and mem to max negative values via msi afterburner help?
    #3
    mthompson2336
    New Member
    • Total Posts : 10
    • Reward points : 0
    • Joined: 2020/03/16 01:56:41
    • Status: offline
    • Ribbons : 0
    Re: 3090 FTW Ultra - six months of driver resets 2021/06/19 21:10:17 (permalink)
    arestavo
    Computer specs help (I'm on mobile, so if you have a signature below the post I dont see it).
     
    With that said, many folks have had crashing issues because their motherboard needs to be updated. Check to see if there's a BIOS update for yours.




    Thanks, I've updated my post with my specs!
    #4
    mthompson2336
    New Member
    • Total Posts : 10
    • Reward points : 0
    • Joined: 2020/03/16 01:56:41
    • Status: offline
    • Ribbons : 0
    Re: 3090 FTW Ultra - six months of driver resets 2021/06/19 21:16:01 (permalink)
    Sajin
    Yes, it does sound like a bad card. Does underclocking the core and mem to max negative values via msi afterburner help?



    I'm noticing that some of the time when I quickly alt-tab to afterburner or x1 it will show the frequency well above my underclock. Underclock is showing, it just seems to be ignoring it sometimes. Not all the time, but not too hard to reproduce either.
     
    I also noticed that if I try and boost lock with an underclock (to try and avoid some transients or whatever) it's not unusual to tab out from a game and see frequencies much higher than the lock, sometimes even higher than the baseline boost frequency.
    #5
    mthompson2336
    New Member
    • Total Posts : 10
    • Reward points : 0
    • Joined: 2020/03/16 01:56:41
    • Status: offline
    • Ribbons : 0
    Re: 3090 FTW Ultra - six months of driver resets 2021/06/20 19:10:43 (permalink)
    Sajin
    Does underclocking the core and mem to max negative values via msi afterburner help?



    Seem to be a little less frequent when underclocking the max in afterburner (-502 core, -502 mem). Today when I run no man's sky it crashes within a minute on normal, maybe 2-3 minutes underclocked.
    #6
    Jstandaert
    Superclocked Member
    • Total Posts : 243
    • Reward points : 0
    • Joined: 2021/04/10 16:36:16
    • Status: offline
    • Ribbons : 2
    Re: 3090 FTW Ultra - six months of driver resets 2021/06/20 20:29:49 (permalink)
    I apologize if I missed the answers to my following questions but I had similar issues (3090 ultra FTW3 made in Taiwan)
     
    - do you have another OC program running or installed you forgot about?
     
    -do you have the switch to the left of 3 power plugs on the card flipped to normal or OC?
     
    -Have you done the CC wipe of the drivers and tried a clean install?
     
    -Do you still have the box? can you verify where it was manufactured?  I had heard the Taiwan ones are notorious for having power issues.   

    Save some Dough-Use my Code
     
     
    #7
    Nozler
    CLASSIFIED Member
    • Total Posts : 2043
    • Reward points : 0
    • Joined: 2009/10/28 10:51:49
    • Location: Ottertail county,Minnesota,USA
    • Status: offline
    • Ribbons : 1
    Re: 3090 FTW Ultra - six months of driver resets 2021/06/20 20:33:32 (permalink)
    Riva tuner installed with After burner? Possible to disable Riva and try again I've had issues with tuner b\4

    heatware
     
    #8
    Jstandaert
    Superclocked Member
    • Total Posts : 243
    • Reward points : 0
    • Joined: 2021/04/10 16:36:16
    • Status: offline
    • Ribbons : 2
    Re: 3090 FTW Ultra - six months of driver resets 2021/06/20 20:55:13 (permalink)
    Nozler
    Riva tuner installed with After burner? Possible to disable Riva and try again I've had issues with tuner b\4


    your sig gave me a stroke. the shadows and large font/small font made me concentrate way harder than I wanted to this evening lol

    Save some Dough-Use my Code
     
     
    #9
    kevinc313
    CLASSIFIED ULTRA Member
    • Total Posts : 5004
    • Reward points : 0
    • Joined: 2019/02/28 09:27:55
    • Status: offline
    • Ribbons : 22
    Re: 3090 FTW Ultra - six months of driver resets 2021/06/20 21:36:07 (permalink)
    Few things not mentioned:
     
    - Vram junction and gpu hotspot temps as measured by HWiNFO.  It can also read ICX temps.
     
    - I usually have the best luck with OSD via RTSS and HWiNFO.  Sometimes I have to set overclock with X1 and then close it out, in particular with RDR2 to prevent stuttering.  However RTSS is known to crash 3dMark benches, so I bench with just X1 and no OSD.
     
    - Can you pass 3DMark or other benches? Unigine? Furmark stress test?  Prime95 AVX?
     
    - Have you tried X1 vs. Afterburner?  Most recent AB?
     
    - What kind of GPU load are you hitting during crashes, are there any FPS limits in place and power limits?
     
    - Have you logged GPU Performance Limit triggers in HWiNFO, reset at the start of gaming?
     
    - Normal or "Prefer Maximum Performance" power settings (can be set by game) in Nvidia driver?
     
    - Windows application compatibility settings and run as administrator?  Disable Windows Defender for game exe?
     
    - Are you not using Geforce Experience? Most people here don't.
     
    - The fact you are not holding underclock is suspicious but not unheard of.  I've not tested it extensively, but have had cases where the curve took but then stopped being obeyed after an hour or so. Turning down power limit, clock and limiting frames seems more reliable.
     
    - Do you have good or bad 3rd 8-pin and PCIe slot power balance?
     
    - 3 separate PSU cables and have you tried replacing the cables and reseating the card?
     
    - Have you tried the normal XOC bioses, the Rebar Bioses or the XOC rebar Bioses?
     
    - What is your definition of not overheating and good case air flow?  Some people think 80C is fine and have all glass cases with a front AIO, have poor airflow even with a bunch of fans.  Back side VRAM could still be getting hot.  What is your ambient room temp and air circulation around the case?
     
    - Ever try another version of Windows?  I've not heard much about the EDU version.  I run the ancient 1809 W10E LTSC.
     
    - Try a different, high quality/OEM displayport cable?
     
    - Change to a different audio device?  Not using one of those wacky Nvidia audio processing utilities?
     
    - Anything in the #2 PCIe slot?  Have you tried the GPU in the #2 slot? 
     
    - How do the VRM/MOSFET temps look on that PRIME board?  The 5950X is pretty power hungry and the X570 Prime is the bottom tier X570 board.
     
    I feel you pain, I had occasional BSOD crashes for 18 months on my main machine until I got a MB bios with better memory handling/auto timings.
     
     
     
     
     
    post edited by kevinc313 - 2021/06/20 21:47:16
    #10
    mthompson2336
    New Member
    • Total Posts : 10
    • Reward points : 0
    • Joined: 2020/03/16 01:56:41
    • Status: offline
    • Ribbons : 0
    Re: 3090 FTW Ultra - six months of driver resets 2021/06/20 23:35:56 (permalink)
    Thanks for the ideas,
     
    Jstandaert
    - do you have another OC program running or installed you forgot about?

    No, I don't normally overclock in the first place, so the only reason I have been trying X1 or Afterburner is to underclock.
     
    Jstandaert
    -do you have the switch to the left of 3 power plugs on the card flipped to normal or OC?

    Normal, although I have changed the fan profiles to be more conservative than the defaults.
     
    Jstandaert
    -Have you done the CC wipe of the drivers and tried a clean install?

    Yes, using DDU. I've also clean installed the studio drivers since I've seen reports that they can be more stable, but no change.
     
    Jstandaert
    -Do you still have the box? can you verify where it was manufactured?  I had heard the Taiwan ones are notorious for

    Yes, made in Taiwan.
    #11
    mthompson2336
    New Member
    • Total Posts : 10
    • Reward points : 0
    • Joined: 2020/03/16 01:56:41
    • Status: offline
    • Ribbons : 0
    Re: 3090 FTW Ultra - six months of driver resets 2021/06/21 00:44:32 (permalink)
    Wow, so many things to consider. Thank you so much for your suggestions.
     
    kevinc313
    - Vram junction and gpu hotspot temps as measured by HWiNFO.  It can also read ICX temps.

    PWR4 is always the highest reaching low 80s. Mem 1, 2 & 3 are equal to each other mid 70s, followed closely by GPU2. GPU1 is the lowest.
     
    kevinc313
    - Can you pass 3DMark or other benches? Unigine? Furmark stress test?  Prime95 AVX?

    Unfortunately, yes. Furmark at 4k was steady at 76 degrees without any crashes. 3DMark Time extreme and Fire extreme both ran without problems. Time extreme stress test passed with 97.8%. Was really hoping to crash out in any of these.
     
    kevinc313
    - Have you tried X1 vs. Afterburner?  Most recent AB?

    I've been updating regularly, currently Afterburner 4.6.4.16094, X1 1.2.1.0.
     
    kevinc313
    - What kind of GPU load are you hitting during crashes, are there any FPS limits in place and power limits?
     

    I've tried both. I normally run with a 60fps limit. Crashes are occurring with loads of 40 or 50%. GPU temp is around 60degrees.
    I have a big monitor so if I turn off limits it can run at 99%, but temp steadies around 70-75 degrees.
     
    kevinc313
    - Have you logged GPU Performance Limit triggers in HWiNFO, reset at the start of gaming?

    Not specifically limits, but have logged sensors with hwinfo64, and at other times with x1 and icue (aoi temps and fans). I don't see anything that spikes or maxes when a crash occurs.
     
    kevinc313
    - Normal or "Prefer Maximum Performance" power settings (can be set by game) in Nvidia driver?

    Tried to change pretty much anything i can, in driver, app override, or in-app. In six months you can cover a lot of permutations :(
     
    kevinc313
    - Windows application compatibility settings and run as administrator?  Disable Windows Defender for game exe?
     

    Will try these tonight, worth a shot!
     
    kevinc313
    - Are you not using Geforce Experience? Most people here don't.
     

    Normally yes with the osd disabled, but I removed it with a driver clean install.
     
    kevinc313
    - Do you have good or bad 3rd 8-pin and PCIe slot power balance?
    - 3 separate PSU cables and have you tried replacing the cables and reseating the ca

    I'm running three separate cables to the PSU. I had fancy corsair colored cables, but switched to the PSU OEM cables in the last few months.
     
    kevinc313
    - Have you tried the normal XOC bioses, the Rebar Bioses or the XOC rebar Bioses?

    I'm not sure what these are, so I will check them out. I have tried enabling resizable bar without stopping the crashes.
     
    kevinc313
    - What is your definition of not overheating and good case air flow?  Some people think 80C is fine and have all glass cases with a front AIO, have poor airflow even with a bunch of fans.  Back side VRAM could still be getting hot.  What is your ambient room temp and air circulation around the case?

    Great point. I like a quiet system, so I'm running an o11 xl with six inflow and four outflow, with an aoi on the top outflow (yes, I went with aesthetics on that one). Fan is horizontal so the three bottom fans flow straight into the gpu. Most fans are ql120s, although I turn off all rgb.
     
    Location seems good, clean uncluttered area in an open floor plan. Fan profiles are pretty conservative, ramping up fairly quickly with a final kick if anything hits 80. With most games the extra flow of all of those fans really helps the gpu stay low (60s), the exception being Ark which tries its best to melt everything.
    kevinc313
    - Ever try another version of Windows?  I've not heard much about the EDU version.  I run the ancient 1809 W10E LTSC.

    The educational version is basically the pro, but with pretty much all the admin extras included. I think the idea is for cs students to get comfortable with the admin tools so they want to use windows in the field. I got it as a free upgrade when taking some courses.
     
    I've been updating the os regularly without any improvements.
     
    kevinc313
    - Try a different, high quality/OEM displayport cable?

    Yes, went with a cable club certified 8k60, 32.4Gbps cable (highly recommended) to drive my monitor. Was having some issues with the OEM cable and this new one was like magic.
     
    kevinc313
    - Change to a different audio device?  Not using one of those wacky Nvidia audio processing utilities?

    I normally run off an external card (FocusRite), but have yanked that and even disabled onboard audio in the bios while trying to eliminate variables.
     
    kevinc313
    - Anything in the #2 PCIe slot?  Have you tried the GPU in the #2 slot? 

    Nothing in the slot. I haven't tried it because bandwidth, but I think I'll give that a shot.
     
    kevinc313
    - How do the VRM/MOSFET temps look on that PRIME board?  The 5950X is pretty power hungry and the X570 Prime is the bottom tier X570 board.

    Was originally a 3900x, only switched to the 5950x to try and solve this. With either, the southbridge gets pretty hot (like all x570s) but none of the other board sensors get above 60 according to hwinfo.
    #12
    Sajin
    EVGA Forum Moderator
    • Total Posts : 49168
    • Reward points : 0
    • Joined: 2010/06/07 21:11:51
    • Location: Texas, USA.
    • Status: offline
    • Ribbons : 199
    Re: 3090 FTW Ultra - six months of driver resets 2021/06/21 01:00:30 (permalink)
    Sounds like you need to test the card out in another pc to see if the issue will follow the card.
    #13
    arestavo
    CLASSIFIED ULTRA Member
    • Total Posts : 6916
    • Reward points : 0
    • Joined: 2008/02/06 06:58:57
    • Location: Through the Scary Door
    • Status: offline
    • Ribbons : 76
    Re: 3090 FTW Ultra - six months of driver resets 2021/06/21 05:58:31 (permalink)
    You posted that your Prime X570-PRO is on BIOS revision 3402? I didn't see that one listed, but 3405 and several newer versions (3 total) including the latest 4002 have system stability improvements.
     
    You might want to update your BIOS and see if it fixes those random crashes like it has for others with AMD boards.
     
    https://www.asus.com/us/Motherboards-Components/Motherboards/All-series/PRIME-X570-PRO/HelpDesk_BIOS/
     
     
    #14
    Gogod2020
    iCX Member
    • Total Posts : 272
    • Reward points : 0
    • Joined: 2020/10/19 14:31:11
    • Status: offline
    • Ribbons : 1
    Re: 3090 FTW Ultra - six months of driver resets 2021/06/21 06:25:34 (permalink)
    mthompson2336
    Sajin
    Does underclocking the core and mem to max negative values via msi afterburner help?



    Seem to be a little less frequent when underclocking the max in afterburner (-502 core, -502 mem). Today when I run no man's sky it crashes within a minute on normal, maybe 2-3 minutes underclocked.




    No man's sky is not an indicative game to judge your GPU with, its very very badly written crashes all the time for all kinds of reasons and for the same no reason at all sometimes it doesn't crash. Usually after restarting the pc NMS works for me but overall it will crash when no other game does and I am playing around 20 AAA titles at the moment so I have a very clear view if my card works or not.

    Intel Core 10700K, Z490 Aorus Elite AC v.1.0, GSkill 32GB 3000/CL14, EVGA 3080 FTW3 Ultra, Corsair HX1200, Dell S2721DGF, DRP4, Be Quiet SW3 140mm pwm high speed x4, Be Quiet 500DX, Samsung NVME 970 Evo Plus 1TB, Samsung SATA 860 Evo 1TB
    #15
    kevinc313
    CLASSIFIED ULTRA Member
    • Total Posts : 5004
    • Reward points : 0
    • Joined: 2019/02/28 09:27:55
    • Status: offline
    • Ribbons : 22
    Re: 3090 FTW Ultra - six months of driver resets 2021/06/21 17:29:26 (permalink)
    mthompson2336
    Wow, so many things to consider. Thank you so much for your suggestions.
     



    Sounds like you've covered a ton of stuff, happy to help.  Forgot to mention the MB bios update yesterday, someone hit on it above, looks like they are up to #4002, probably worth a try. 
     
    Also (long shot) messing around with the number of cores running (like disable half) and turning off things like speedstep and hyperthreading, or the AMD equivalent.
    post edited by kevinc313 - 2021/06/21 18:15:57
    #16
    Jump to:
  • Back to Mobile