cancel
Showing results for 
Search instead for 
Did you mean: 

Graphics Cards

Toast21
Adept I

6900XT causing PC to reboot while gaming

Ever since I upgraded my gpu from 2080super to 6900xt I get random reboots when gaming. I can be fine for 3 hours and it will reboot the system without warning. At first I thought it was the psu (750w Corsair) but I upgraded that to a 1000w Corsair psu and same problem. Gpu temps stay below 85c and cpu stays below 55c. When I had my 2080 I never experienced such problems. I deleted the nvidia drivers before installing the 6900xt too.

msi z490 gaming edge wifi 

i7 10700k

corsair rm1000 psu

gigabyte 6900xt

32gb Corsair vengeance 

Any help? I don’t know what else to do to trouble shoot this, I get no warning when it reboots, when it is running it runs great. 

187 Replies

Yeah I've never seen over 500W at the wall running games and 3Dmark with my XFX 6900 XT MERC BLACK. I wouldn't pay too much attention to reddit.

I'd be happy with an 850W PSU but I run 1000W anyway. I'd also run 3 cables just because, however I do agree 2 and a daisy chain would be fine with this power draw.

I also don't think a lot of people are having issues with graphics cards in this thread. One guy with lines on his screen, yeah he most likely does. The rest could be many different things that are hard to narrow down.

For me I've not been able to reproduce it since the AGESA 1.2.0.3a BIOS update, and the issue has occurred less and less throughout the year as I installed newer drivers and BIOSes. I've also ruled out GPU and PSU.

How many people have tried their graphics card in a different PC and had the same issue?

I'm glad I found this post because I've been getting this issue right now on my PC and only happens on certain games. I use the DDU method in safe mode tried multiple amd radeon software versions same issue. I thought it was my power supply as I had a 850w platinum psu that's now in my son's build working flawlessly with his 6800xt and I bought a new 1000w Titanium psu and still having random reboots when gaming and is super annoying. I'm thinking of buying a 1200w PSU, do you think that will fix the issue? 

0 Likes

@N8iveWarMachine check out the post from Matt. You've been on 2 PSU's I think it's safe to say you've ruled out a PSU issue, an 850W would be fine. Time to rule out other components, could also be a software issue.

However many have been replacing their CPU and fixed it. I ruled out everything except CPU, MOBO, RAM and SSD. I replaced my CPU several days ago and no issues since and something about the way it runs and loads the cores differently gives me a gut feeling it's fixed, however far too early to know for sure.

Rule out what you can, make sure you update everything BIOS, chipset drivers and all that stuff, maybe try a fresh install. Once you've narrowed it down the best you can, you're left with some parts that potentially need replacing.

While my small local PC store was testing my CPU another person brought in another CPU with the same issue. This doesn't mean in your case it's definitely the CPU but it's something to look closely at and rule out.

 

0 Likes

Solved for me: A Bios reset fixed my issues.

 

0 Likes

Solved for me : I bought a 3080 from EVGA in Hybrid which sometimes consumes more than 400Watts in overclocking and ... more NO PROBLEMS by having only changed the graphics card ...

Wanna hear something funny? ( for you not for me)

After 2 weeks of no crashes i changed to a new case and guess what, i had another crash. What does the new case have that the otehr doesn't ? More fans.

This is where AMD loses most of its clients, their soft is garbage. Excuse my agressiveness but when you pat 2000$ for a **bleep** silicone you expect it to "Just work", not run around and around like ian a **bleep** circus until something( most of the times you don't even know what) fixes your problem, if it ever.

Shame on you, AMD

 

Are your games crashing or entire Windows will reboot? Is your Eventlog filled with WHEA unrecoverable errors just before crash? When i have moved from 1080p to 1440p gaming, my memory become unstable at 3600 Mhz and computer has started to reboot. Undercocking memory to 3200 Mhz has solved my problem, but of course it can be different case with you.

 

My whole story: Solved: Random reboots after monitor upgrade - AMD Community

I've not seen WHEA errors, but I might not be looking in the right place. I believe my Corsair memory is 3200 anyway.

I've had some minimal feedback from Scan, too, and they've got it working on a 4K monitor with the latest BIOS, but that is not the 1440p 60Hz monitor that I run at home.

I would ask if anyone believes that running the benchmark (3dMark Time Spy) at 1440p 60fps but rendering it at 4K would likely be a different thing for the GPU/CPU/MB/Mem to manage than simply running at 4K straight.

I do know that 1440p is not always perfect ¯\_(ツ)_/¯

---

@x6205I've read your story now ... you shouldn't have to limit your PC ... it should run at stock levels ... it's not acceptable.

0 Likes

Well, i understand your point of view, but why should i stress myself and spent whole summer with troubleshooting. Tinkering with BIOS or DRAM Calculator or blaming AMD for bad drivers, Asus for bad BIOS or AMD for bad AGESA or whatever. Memory downlocking was propably easiest way how to solve my specific problem.

Other option i have is replace RAM with even more expensive model and if it will not help then what? Maybe it's bug in AGESA, maye in drivers. Since then i have updated BIOS so maybe i will re-test it at 3600Mhz again. Anyway, difference between 3200 vs 3600 Mhz is so small, that it's almost irrelevant to push RAM higher. 

3200MHz vs. 3400MHz vs. 3600MHz vs. 3800MHz vs. 4133MHz | Best RAM for Ryzen 3000 - YouTube

@Cmdr-ZiNI left my PC on for 3 Days on a fresh install, had no issues in Windows 10. However, that was the Basic VGA Driver. The moment I installed AMDs proprietary wares... GPU Locked up, or Blank Screens. I could see quick tempered people going NUTS at their builds.

 

I've been on Linux since June 2021. I've had no issues with this card. I decided to dual boot into Windows 11 when it was released. I have to say that DCOM thing is NOT working very well on the new OS. I tried watching a video and the moment I skipped a few scenes or fast fowarded n my screens all went blank.

It's software, it has to be. I've been running fine on POP OS a variant of Ubuntu Linux, completely normal and the Basic VGA ran normal without fault.

Everyone seems to be racking their brains out thinking it's hardware. I did even think that @RaiJinRoh  PSU or Power Requirements had something to do with it. I have a 1000w PSU and an i9. My Nvidia card worked fine, however AMD cards are renowned for being very power hungry like their CPU counterparts.

But no issues on Linux. - Are these forums monitored! Like is there anyone from AMD that actually reads these forums? What do you have to do to get the attention of a mod or AMD Developer? Post a nude?? These forums are annoying and several people have contacted AMD including me and they are refusing to acknowledge the problem. 

I did checkout TenForums, a few people were complaining that Windows Sucked and kept rebooting on them, after I asked what GPU card they were using 100% of them said AMD.

https://www.overclockers.co.uk/forums/threads/the-radeon-rx-6800-xt-owners-thread.18906578/page-35  - There is also a thread on OverClockers

https://gpuopen.com/  I've had a look here.

https://github.com/GPUOpen-Drivers/AMDVLK 

 

I'm hoping someone else that's fairly tech savvy or has developer/software engineering skills, can test out my theory and install a copy of POP OS Linux on perhaps a spare hard drive or partition or even a USB stick install.

One thing, I did notice however on Windows, if I disabled my wallpaper and all visual effects I wasn't plagued by the reboots. Whilst on idle. I do remember that from Windows XP, Microsoft changed the way the Graphical User Interface was drawn by the OS and on Linux the UI is less graphically enhanced, depending on which GUI your using as Linux has a fair few choices.

 

Does anyone have a PSU higher than 1000w? and still plagued by the problem.

I suppose another option would be to port the Linux drivers over to Windows, but Windows Requires Signed Drivers. You can override that, I guess. But how many security overrides must you enable just to use a graphics card. It's like your literally fighting Windows's Subsystems.

I know reverse engineering and publishing the source would be illegal but porting the Linux drivers seems like feasible option, though the interface architecture is completely different. Although being sued by AMD, would at least put me face2face with AMD.

https://www.overclock.net/threads/official-amd-radeon-rx-6800-xt-owners-club.1774300/page-9  - Some people here are suggesting to reduce the power consumption of the card by under clocking it.

 

One thing does baffle me. Not everyone is affected by this. Some cases, the system runs fine. Makes me wonder if it's a GPU BIOS flaw. As different manufacturers have different BIOSs https://www.techpowerup.com/download/ati-atiflash/ 

List OF GPU BIOSs

https://www.techpowerup.com/vgabios/?architecture=AMD&manufacturer=&model=&interface=&memType=&memSi... 

 

Although if your card or system crashed or hung during the update. Unless your card has a dual BIOS, that would be the end of that card, unless you wanna resolder a new BIOS chip to the card, though you'd need some master skills for that feat of engineering.

@Tazryder 3 days may not have been enough but depends on the usual frequency of the issues you're having.

You say you have an i9 CPU so you probably don't have the random reboot issue many Ryzen users are having. Mine I'm putting down to CPU not being as stable as others, since replacing it I've had no issues for just over 2 weeks running constantly and I hope I don't again. My memory has been pulled from the QVL listing I wonder if that was a contributing factor.

Yours sounds more like a GFX card issue, it could be software. I doubt it's power draw, 1000W is fine assuming your PSU is in good condition and setup correctly, I assume you're running 2 separate power cables I can't remember all the previous posts. AMD graphics cards use a lot less power than their competition this gen BTW, but it's not a power issue.

If you want help from AMD the forums aren't the the best place it seems, email their technical support.

It might seem like everyone has the same issue but there's lots of causes and people on the internet can miss things so you need to take what you read with a grain of salt until you can confirm it yourself.

If there was a mass issue there might be a statement but it's more likely lots of different issues that would need individual technical assistance and  forums aren't the best place for that.

When you installed AMD's wares what did you install, was it driver only? I have had the drivers glitch on install had to DDU and resinstall and they were good since.

Do you have any overclocking software installed like MSI Afterburner?

As long as your drivers and Windows and everything are up to date then I'd contact your graphics card manufacturer and ask them.

If they recommend a BIOS update I'd do it. A long time ago I messed up a BIOS update on a GFX card, I was able to boot with another graphics card and flash the messed up one in windows, I doubt this will work these days you'd probably need to RMA.

However these days as long as you follow the instructions to the letter a BIOS update is pretty easy to do and very reliable as long as there's no power outage you'll be fine. Check with the GFX card manufacturer first.

I wouldn't be surprised if you need to reinstall or update something or need to RMA your GPU. It's hard to say the cause without looking at it more closely.

Some of the suggested fixes here, while they might have seemed to work for some people don't appear to be getting at the root cause.   I installed my card, things generally worked and then I decided to run 3dmark Time Spy to see how it would do.  After only 10-20 seconds in, the system shut down.  And when I say shut down, the red Radeon logo shut off, but the LED's on the RAM were still lit.   So it wasn't a complete power shutdown, as it turns out it appears to have been the 12V supplies.  My PSU is a Corsair HX750i and should be plenty for what I have installed in the machine. It's got a Ryzen 5800x at stock frequency/voltage, 32GB 3600 DDR4 and a 2TB PCIE3 M2 SSD.   None of that should come anywhere close to the limit, even if the card were to draw 500W.  Then I read a comment somewhere on Reddit about split rail supplies.   In my case, I can select this in the iCue software.  When I loaded that up, sure enough I had it set to split rails. (not intentionally)    So with split rails, that means the over-current protection limit for each of the two rails will be lower.   The problem is, I have my power connectors to the GPU coming from one of those rails.   Once I rebooted with that change, the shutdown problem was resolved.   I'd suggest that people check this on their PSUs.   So no, you don't need a 1000W+ PSU, you just need to allow it to supply the power it needs on one rail.

@sebakke yeah you want to use 2 separate power cables for your GFX card and put one on each rail. I'm pretty sure this was mentioned earlier.

Although your stability may of improved keep an eye on your system closely over the next couple of months because there's instability issues with the 5000 CPU series too and they sound a lot like how you described.

I wonder if there's issues with memory speeds over 3200 Mhz or the system running the fabric too high. I too have a 5800X and 3600Mhz RAM, infact it's my second 5800X it solved my issues perfectly for 2 months before also dying.

There's something unstable about this platform.

0 Likes

My 5800x has been perfectly stable since I built it 9 months ago.  I've used it with a 5600xt and 3060 without any issues.  But either of those GPUs are relatively low power, so haven't stressed the system.  

0 Likes

@sebakke that's good to know you where stable with a 5800x CPU and 3600Mhz RAM prior to getting your graphics card.

Maybe I'm just unlucky to get 2 dud CPUs, but it does make you wonder, especially when a lot of people say Ryzen has troubles with high speed memory.

0 Likes

Hi everyone, I have the same problem here but with an Asus TUF 6800xt. So while playing Remnant my screen will turn black and restart the system. It can happens after only a few seconds to up to an hour.  I then have to press the power button to have my system back, a simple reset won’t work.  So here is my system:

 

Mobo: Asus TUF X570 WiFi with latest bios (4021)

Ram: 16Gb Gskill Trident(3600)16gb and 16gb Corsair Vengeance (3600)

GPU: Asus TUF 6800Xt

CPU: Amd 5800X

Windows 10 Pro latest build fully updated

 

Keep in mind that before putting an 6800xt about a few months ago in the above system, I had an 1080Ti with zero issue. So here what I did to test the stability of the system:

 

-   Prime 95 about 8hrs no issue

- OCCT(Cpu stress test) 8hrs no issue

- Time Spy stress test no issue

- Unigine Superposition stress test 4 hrs no issue

- Furmark 2hrs no issu

- Window Ram test no issue

 

Now the first thing that I tested was to use only one memory brand at a time. I even underclocked them to 3200 instead of 3600 but same problem.  Clean drivers installation with DDU, same problem. Changed my Seasonic 850W gold to an EVGA 1000W gold, same problem. Plugged my GPU with two distinct cables instead of a split one as suggested, same problem.  Used different Amd drivers and also updated to the latest one(21.9.1), same problem. SAM on or Off, didn’t make any difference, same problem.  I also tried to manually install the driver without the Amd Software, same problem.

 

I used Amd overlay to monitor the GPU. So here what I noticed while playing. In demanding part of the game where the wattage will get high the system will/could crash. It seems that a sudden spike in wattage can cause the system to crash.

 

I have stressed test the GPU with many software with zero issues.  In Furmark for example it will draw up to 300W and the junction Temp will reach a temp of 92C. But then again, when playing games I will rarely reach that power draw nor that I would reach those temps, but wattage will greatly fluctuate.  The fact that the power draw is stable in stress tests may explain why I don’t crash.   

 

  For example when I play Soul Calibur 6, I will never crash because the game locks the FPS to 60 and the game will rarely draw more than 100W. But in Remant with uncapped FPS my GPU wattage could go anywhere from 100W to 280W. I believe that some odd wattage spike may cause my system to crash.

 

So what I did to be able to play Remant was to underclock my max frequency from 2339 to 1900 to limit the power draw. By doing this, zero problem, no more crashes.  I’m reluctant to RMA my card because I don’t think it will fix the problem and it doesn’t seem to be a hardware issue. As stated I have zero issue while stress testing. I’m not sure it it’s a drivers, bios or firmware thing but it needs  to be fixed otherwise  I may have to sell my 6800xt and to replace it with an RTX 3080.

 

0 Likes

For now i didn't get any more crashes.

I have noticed that i didn't disable onboard sound in Bios after the last Bios update. Lets see if it continues for a week. The problem i had was getting the pc to restart when i was entering Lost Ark game or in Horizon Zero from time to time. Another error i was having was total lost of signal when i was activating HDR. (My hdmi 2.1 cable is a good one ) 

Hi, RaiJinRoh, a few questions to just clarify for the good of the thread ... you really don't have to answer them, but if you can then hopefully it will assist others in their fault finding:

  1. Which motherboard are you using?
    • If it is the same as mine (ASUS X570-E see previous) are you on the latest 3604 (from 2021/05/07) BIOS or the Beta 3801 (2021/05/07)?
  2. Did Lost Ark or Horizon Zero crash whilst you were playing the game?
    • If so, can you replicate exactly where repeatedly?
  3. Do you own 3dMark and the Time Spy benchmark (my selfish question!!! ) you can test with?
  4. Are you running either in DirectX 11 or DirectX 12?
  5. What power supply (PSU) are you running?
  6. When you say "total lost of signal when i was activating HDR" is that the system shutting down, or is it just display on the HDR monitor?

Sorry for the slew of questions, but every little bit helps.


@RaiJinRoh wrote:

For now i didn't get any more crashes.

I have noticed that i didn't disable onboard sound in Bios after the last Bios update. Lets see if it continues for a week. The problem i had was getting the pc to restart when i was entering Lost Ark game or in Horizon Zero from time to time. Another error i was having was total lost of signal when i was activating HDR. (My hdmi 2.1 cable is a good one ) 



EDIT - Oh, @milkboy31 / @RoboCop / @Lego_Underfoot (and you, RaiJin), I did actually check out a few other threads in other arenas. Some folks are reporting similar issues in the CPU section in the following thread. This reply, in particular, speaks about a recent Beta BIOS regarding some AGESA settings. Could this be the magic bullet?
https://community.amd.com/t5/processors/ryzen-5900x-system-constantly-crashing-restarting-whea-logge...

0 Likes

  1. MSI B550 A PRO and i am on the newest Bios now ( not a beta anymore since a week back )
  2. It crashed randomly sometimes when i entered the game sometimes while i was playing ( my temps are ok). Noticed that in horizon zero it did crash more time when a cutscene was starting.
  3. I don't own any of them but in furmark the stress test was ok. I have tested the benchmark in horizon zero down also and worked fine.
  4. I run with Directx 12
  5. Seasonic Focus Gx - 850  ( 850w Gold) 
  6. Just display on the LG OLED TV

 

Other things that i have changed recently

  • cpu to run at 95% ( donno why i did that as my temps are ok)
  • i have the SAM on, it was off since last Bios but i was getting crushes before also
  • dissabled dcom to see if it help

The problem with Horizon Down might be more from the game than from my 6900 XT the game has huge memory leaks and was eating all my 16GB of VRAM, almost, on 4k. I changed to 1440p and works perfectly.

I let you guis know if this is totally fixed.

BTW. My 6900XT its underclocked/undervolted so it might change the solution a bit.

I have gigabyte 6900xt and keep it at 2400MHz, 1125V, max power (15), no memory change. The GPU crashed on Auto setting also so it might have nothing to do with that.

 

 

Thanks for all that information, mate ... I'm 100% sure it'll contribute going forward.

Can't explain those CPU things, but I'd be curious if your MSI B550 BIOS update was addressing AGESA stuff, like the linked thread above.

---

FWIW - I think your undervolting is probably really helping you out, yeah, mate.

None of us should have to *UNDER* power our brand new top of the range graphics card, though ... At a bare minimum it should work just fine at stock levels, right?


@RaiJinRoh wrote:
  1. MSI B550 A PRO and i am on the newest Bios now ( not a beta anymore since a week back )
  2. It crashed randomly sometimes when i entered the game sometimes while i was playing ( my temps are ok). Noticed that in horizon zero it did crash more time when a cutscene was starting.
  3. I don't own any of them but in furmark the stress test was ok. I have tested the benchmark in horizon zero down also and worked fine.
  4. I run with Directx 12
  5. Seasonic Focus Gx - 850  ( 850w Gold) 
  6. Just display on the LG OLED TV

 

Other things that i have changed recently

  • cpu to run at 95% ( donno why i did that as my temps are ok)
  • i have the SAM on, it was off since last Bios but i was getting crushes before also
  • dissabled dcom to see if it help

The problem with Horizon Down might be more from the game than from my 6900 XT the game has huge memory leaks and was eating all my 16GB of VRAM, almost, on 4k. I changed to 1440p and works perfectly.

I let you guis know if this is totally fixed.

BTW. My 6900XT its underclocked/undervolted so it might change the solution a bit.

I have gigabyte 6900xt and keep it at 2400MHz, 1125V, max power (15), no memory change. The GPU crashed on Auto setting also so it might have nothing to do with that.


0 Likes
gbo3000
Journeyman III

I had the same problem with 5700xt. specifically it happened when playing Scum. tested with OCCT(PSU) and it crashed instantly.

tried all suggestions still not work. So i decided to use two separate pcie cable to connect the card and ran test again. Now like magic it doesn't crash when running OCCT.

Let me know if this works.

0 Likes
sherox888
Adept II

Make sure the power supply is safe. I had the problem on the 1000 Watt EVGA G3 Power and when I returned to the EVGA G2 Power. 1000 watts The problem is gone, as well as make sure all the power connections are in their proper place. If you have a friend, try his psu.

@sherox888 He's already tried 2 different PSUs, I seriously doubt it's his PSU if he's run the cables correctly. I'm betting he has a Ryzen CPU, but there's other things that could cause this.

0 Likes

Okay, sorry for the late reply. After having my gpu in California for the last month getting replaced I finally have a working PC. I ran my PC with furmark for 10 minutes, which it crashed within 2 minutes. I can't say it's an easy fix as I got a nitro+ 6900xt vs my reference 6900xt, bought a new motherboard (because my 16x slot was broken), and updated my bios and installed a fresh version of windows. After all that my computer now works. I don't know if it was the bios update (a new one came out within the month I didn't have a working computer), the drive wipe, or the new motherboard that fixed the issue, but the issue is now fixed on my system.

0 Likes

Small update on my issue (because getting detail on it has not been easy) and that is that there is "kind of" no update.

Basically, I sent the PC (with my case AND 1000W PSU - which weren't part of the Scan build) but not my drives to the company that made the build, and they have not been able to replicate the issue that occurred literally every time I ran the Time Spy benchmark.

In my gut I believe that the incident that (for those genuinely getting it) is happening to everyone isn't actually individually related to the Graphics Card ... and maybe not the CPU, either ... but more to how the systems themselves are handling sudden power requests through the MB / BIOS.

The BIOS updates that we've been seeing keep speaking about USB problems, and there's ZERO mention from AMD in any communications anywhere acknowledging that the issue that is being seen here (and I believe across multiple, differently titled, threads here) is related to either;

  • their hardware operating with strange methods,
  • their BIOS / software handling the hardware requests.

However ... I wonder if any of us have attempted to check USB in any of this? I think (for my part) I know that I'm pretty sure I've had any and all USB 3 hubs plugged in each time ... which ... *maybe* it's a thing? I dunno.

Either way ... I'm asking if I can have a video call with the engineers at my retailer.

@eliotcole the closest I could find that they're aware of the issue is https://www.guru3d.com/news-story/amd-agesa-combo-pi-v2-1-2-is-coming-for-msi-500-series-and-400-ser... they talk about power management changes and trying to fix WHEA errors with BIOS updates.

I suspect it's to do with load line calibration or balances and with infinity fabric and ram over 3200Mhz it just becomes too unstable. Also adding more PCIe gen4 items I suspect makes the stability worse. This coupled with a CPU that is just barely stable and you'll have issues.

This kind of issue has been around since Ryzen 3000, however not everyone will have the same cause, lots of things can cause random restarts. I'd research Ryzen random restarts and if your issue seems similar I'd rule out your CPU.

I've been running my new 5800X for 10 days straight now, Zero issues. I could always reproduce the issue in less that 6 or 7 days before, the only thing I've changed is the CPU, nothing else.

Cheers, man ... I mean, what *is* a positive response from the 'no real update' was that they indicated no issues with the PSU.

We all suspected there wouldn't be, but I dearly didn't need to have that be an issue on top of this issue. Although I hold my hands up to not having tested using purely modular cables on the GPU (not possible for the MB+CPU), I will admit to my brain being a bit on the wrong side of things there.

Still, I wasn't a fan that they literally didn't test the thing that has (without fail in my testing) caused the incident. I respect that they have their process, but they said that they couldn't do a video call with the testing (like, why?) and they hadn't tested the reproducable events that I could do in a second to prove the incident. I still stand by Scan as an excellent company, but this particular experience has been sub-par in general (with some great service from certain staff, though).

I've asked them to check how many ranks the ram has, whilst it's in there (as I can't get the exact SKU and details for some reason), because thanks to Linus it's another thing I don't wish to be losing performance from.

I also respect that from AMD's perspective, a whole new way of managing how memory/disks/GPU memory is handled in game is going to have some way to go before it beds in nicely.

Matt_AMD
Community Manager

If you are experiencing stability issues with your 6900 XT which causes your system to reboot unexpectadly, please refer to and follow this troubleshooting document which if followed step by step, should help you narrow down and identify the cause of your issue.
https://www.amd.com/en/support/kb/faq/ts-tips

If after following that document and applying all of the recommendations you continue to experience the same issue, please open a new discussion and provide the information required below so that the community can offer suggestions.
https://community.amd.com/t5/knowledge-base/information-required-when-posting-a-discussion/ta-p/4227...

Alternatively, please open a e-ticket directly and provide the information requsted in the link above.
https://www.amd.com/en/support/contact