Experience a sudden shutdown to a black screen on your RX 7800XT.
Experience a sudden shutdown to a black screen on your RX 7800XT.
This has been asked, reported etc. multiple times not only here but all over the internet most posts are about a year old, I am hoping some Tech God will return to me with a solution that has been found but the old posts don't have it reported. Anyway so, playing any game (tested RDR 2, CS2, Marvel Rivals) at any point 30min to 2h of the session the screen will go black, but the apps will still be working ill hear audio, I can talk to my friends on discord, but to get back i need to perform a hard reset. After restarting 9/10 times i will find i have no FPS this is because the driver for the GPU fails to start. I must uninstall the device and check for hardware changes (all in device manager) and i will be able to play again for the time being until another crash happens. Since the pc is brand new, I did not want to undervolt, underclock anything fearing it would void warranties and ill be told to piss off by the manufacturer or my PC shop. So I only tried the basics of the old posts like: Turning off SAM (Smart Access Memory) Turning off (or adjusting) various windows settings like power and turning off fast boot) The temps are good about 80c for both when you reach extended gaming. BIOS is updated. I've noticed both in adrenalin and MSI after burner i find the the core clock (max?) is set to 2655 MHz which is already 135MHz more than the manufacture states (Boost clock up to 2520MHz) and during my testing it goes well above even 2655MHz at times. And the power draw... Well i cant find the TDP of my Asrock BUT it definitely spikes to 300W and above at times. when the reference TDP is 263W I think. PC (assembled and "tested" by the computer shop) not a prebuilt i picked the parts myself: AMD Ryzen™ 5 7600X MSI MPG A850G PCIE5 - 850W 80 PLUS Gold (up to 90%), modular, PCIE5, ATX 3.0, 12VHPWR connector Cooler Master | HYPER 622 HALO White | Cooler Kompiuterio(-ių) surinkimo paslauga ASUS PRIME B650M-R Micro-ATX ADATA LEGEND 800 1TB PCIe M.2 SSD Deepcool CC560 V2 PATRIOT VIPER VENOM DDR5 32GB 2X16GB 6000MHz ASROCK RADEON RX 7800 XT Steel Legend 16GB GDDR6 256bit Drivers: Full Adrenalin not just driver 24.12.1 Windows 11 Pro 24H2 build 26100.2605 Chipset6.10.17.152 I play FPS mostly so i run a 1080P monitor with 144hz DP I cannot see the crash in Event Viewer but i was "confirmed" it was the GPU after Marvel Rivals on launch said they detected a GPU crash. Some people said the 12V bus should provided atleast 50 something amps well this straight from MSI I don't think it's the PSU If I would've known this was an issue i would have gone with another manufacturer to be frank, but I am coming off the AMD CPU hype about how they are the greatest, so i though the same about the GPU's. This is my first real pc and it is sort of a "dream come true" for me. So I find it really frustrating that this system is brand new and still unstable to the point of being a paper weight. I've seen others "fix" this issue by getting a 1000w PSU but that is ridiculous my GPU manufacturer states it recommends a 800w PSU, I'm not running some X3D high performance CPU the system should work, and my PSU is 850w gold. This is a midrange card with a 1000w PSU you're getting into 4090 territory. This is also posted on reddit AMDHelp there i've been instructed to disable EXPO and run a benchmark then test if I crash again, that is currently in progress. But i also dont want to disable EXPO as then i'm not taking advantage of the DDR5 speed... I would expect the system to work properly since its brand new had a BSOD today in the morning aswell
Timeout Detection and Recovery (TDR) is a Windows capability that identifies graphics cards experiencing response issues. It can temporarily halt and restart operating systems. The feature resets GPUs when compute kernels exceed a certain duration. To prevent unintended device resets during extended operations, raise the TDRDelay setting. Launch the Run dialog by pressing Windows + R, then type regedit and press OK to access the Registry Editor. Navigate to HKEY_LOCAL_MACHINE\System\CurrentControlSet\Control\GraphicsDrivers. If the TdrDelay key is present, double-click it and adjust its value in the Value data field (e.g., set to 20 seconds or more). Proceed to Step 7. If the key isn’t found, click an empty area on the right pane and select New > DWORD (32-bit) Value. Continue to Step 5. Label the key TdrDelay, double-click it, and input a suitable value (such as 20 seconds or higher). Click OK to save. Exit the Registry Editor and restart your computer for the updates to apply. Additionally, consider disabling MultiPlane Overlay MPO.
Hey there, friend. I understand what you're going through. This situation has happened to me too. A black screen with background noise is often a sign of a driver problem. When you restart and type 'Event Viewer' into the search, open it and look under Windows Logs > System. You should find entries around the time your PC froze. Those logs can help pinpoint what went wrong. Once you have that info, the next steps become clearer. Here are some suggestions: 1) Use DDU to reset your GPU drivers and reinstall them. 2) Perform a clean install of Windows 11—don’t just upgrade or repair. This usually fixes software glitches. 3) If you have an overclocked CPU, GPU, or RAM, try running everything at default settings. Adjust one component at a time to see what helps. 4) Test in a different PCIe slot inside the case. 5) Check if another PSU causes the issue—sometimes power supply problems are the culprit. 6) See if the problem appears on another machine. In my experience, a bad PSU was the main reason for the black screen after installing the GPU.
I repeated my apologies several times, realizing I didn’t mention it clearly. I’m also considering setting up just the driver without any rush. 2. I plan to revisit this later because testing can be slow—sometimes crashes happen in a few hours. I’ll save this for now. 3. Right now everything looks standard. I went to the default BIOS settings and turned off EXPO. 4. I can’t test it here since it’s an MATX board with only one PCIe slot for the GPU. 5. This PC is my upgrade from a laptop, so I’m not sure about the issue. 6. Maybe a friend could check if assembling it helped; they might also test the power supply. Also, I asked my girlfriend to share some screenshots. The CPU seems overheated, and the main problems are consistent with the crash reasons. There are other odd errors, but none link to graphics or drivers. As of now, the test finished okay with a lower voltage GPU and EXPO disabled, ready to try again later.
Nice one. Might be a long-shot bit a good old Windows reinstall is generally a good early win. Do you have any 'Warnings'? These Critical errors will be it registering you pushing the power button rather than doing a clean shutdown. Don't worry about these, I've got loads of them as well. Yeh, 95C does seem hot. Probably within spec but a little toasty. Have you re-seated the cooler and reapplied thermal paste? I've only recently jumped ship from team blue to red and waiting for my new 9800X3D to arrive so don't have much AMD experience yet. So, I did this which seemed to help for a little while but it wasn't long before my black screens returned. Personally, I would take the PC back to where you bought it. If is relatively new, it should be under warrantee and if the issue is easily replicated then there shouldn't be any problem with them diagnosing and finding a solution.
Thank you for everything Sir, I wont pester my girlfriend with error logs anymore ill check them when i get back. The temps on test for me seem expected because i ran extreme preset. when gaming it doesnt go above 85. Im waiting for their email response if they dont respond today im taking the pc to them tommorow (unless i find a fix) still need to try 191x7 advice
The issue stemmed from RAM compatibility with EXPO, as noted by @ChrisLoudon. The RAM module wasn't listed in the Qualified Vendors list for the motherboard, which means it couldn't ensure stable performance. After turning off EXPO, everything worked fine, so I reactivated it and ran a memory test. Both EXPO versions (6000MHz and 5600MHz) crashed within a minute with over 10,000 errors, whereas the non-EXPO setup functioned properly. Evidence attached for reference.
Talked to them really quickly. After a long time it happened once more. Probably lasted even longer.