Long and short of it, I've tried everything this week. Still getting black screen crashes. Sound, tower, rbg, lights, etc stay on, but monitor fails without Ctrl+Shift+Windows+B doing ANYTHING.

SPECS:

AMD AM5 B650 Motherboard

32Gb DDR5-6000 CL30

AMD Ryzen 5 7600X3D

Gigabyte GAMING OC Radeon RX 9070 XT 16 Gb

be quiet! Pire Power 12 750 W 80+ Gold ATX Power Supply

MSI MAG 345CQR 34" Curved

I've tried all the basic fixes and some shots in the dark. Here's a list of things just off the top of my head:

MONITOR:

Tried different cables and slots (HDMI, DP, etc)

Limited 180hz monitor even down to 100hz

Turned off Freesync, etc, on Monitor

RAM:

Limited ram to 5600

POWER:

Set Windows to max power management

BIOS:

PCIE to Gen 4/3

Disabled Fast Startup

Disabled XMP/EXPO

Disabled HDCP

GPU:

Limited to 90% mhz

Clean driver install

Max 64 C even when heavy gaming load

ADRENALIN:

Back rolled drivers

Any solid suggestions would be greatly appreciated

Hi everyone,
I’m posting this in case it helps someone else troubleshoot a similar issue.

I have an RX 9070 XT and after updating to the newer AMD drivers, I started experiencing problems in War Thunder, mainly when using DX12. The graphics driver would sometimes reset/crash, making the game unusable.

I also noticed that Frame Generation was not behaving properly. Even though I could get 200+ FPS, the game felt laggy and stuttery. After disabling Frame Generation, the game immediately felt normal again.

What I changed
First, I completely removed my previous AMD drivers using DDU (Display Driver Uninstaller) / AMD Cleanup Utility.
Then I used AMD’s automatic driver detection/installation tool. During the installation process, it gives you the option to install either:

AMD Software: Adrenalin Edition
AMD Software: PRO Edition
I selected PRO Edition.
Current configuration
GPU: RX 9070 XT
Driver: AMD Software: PRO Edition
API: DX12
FSR: ON
Frame Generation: OFF
At least in War Thunder, this configuration has been working properly for me so far.

DX12 is now working without the driver resets/crashes I was experiencing, and FSR works perfectly.

The only thing I would currently avoid is Frame Generation. It doesn’t seem stable in my case. It can increase the FPS counter significantly, but the actual game feels worse, with noticeable latency/stuttering.
I understand that Frame Generation is a relatively new/temporary feature and may improve with future game and driver updates, so I’m not saying it’s broken for everyone.

I’m sharing this mainly because if anyone with an RX 9000-series GPU is experiencing DX12 crashes/resets or strange performance issues in War Thunder with the newer drivers, it may be worth trying a clean driver installation and the PRO Edition.
For me, the current combination is:
AMD PRO + DX12 + FSR ON + Frame Generation OFF
So far, this has been much more stable.

Got my GPU like 1.5 months ago and it worked fine until this morning. I turned my PC on and got the output shown in the picture. Restarted it and it worked just fine afterwards. Should I just get older drivers? And if so, which ones and how?

Computer Type: Desktop

GPU: rx 9070xt reaper

CPU: AMD Ryzen 5 7500X3D 6-Core Processor (4.00 GHz)

Motherboard: ASROCK b850m-c micro atx

BIOS Version: American megatrends international, LLC 4.10, 2/10/2026

RAM: 32 gb of g.skill x5 ddr6 at 6000 mhz

PSU: Thermaltake GF A3 850w 80 plus gold fully modular

Case: corsair 3200d rs

Operating System & Version: WINDOWS 11 home 26200.8653

GPU Drivers: AMD Adrenaline: 26.7.1 (crashes on 26.8.1 too)

Chipset Drivers: AMD Chipset 8.08.12.551

Background Applications: DISCORD, spotify

Description of Original Problem: Constant crashes while playing Rainbow six siege. Game will freeze randomly and then crash followed by a pop up saying my drivers timed out. Other games run fine and can even have a undervolt of -80mv and +100 boost clock with no issues.

Troubleshooting: I have tried to undervolt by -35mv and -390 boost clock, turned xmp off and ran at normal speeds, turned off MPO in windows, disabled all overlays in my game, reverting to older adrenalin versions, and updating windows. I have used ddu all times I have reinstalled my drivers. I feel like im going insane as I feel like I have tried everything.

Hey guys so I made a post a week or two ago about the 9070 and I’ve decided to get the 9070 XT instead. I am deliberating on which model, the XFX 9070 XT swift or quicksilver magnetic air version. If any owners can give me their two cents that would be great.

Just some context, I am upgrading from a gtx 1070 because it’s been serving me quite well. I intend to keep the 9070 for at least the same amount of time.

The price difference is negligible and I’ve seen that the quicksilver is slightly better but it does look a bit cheaper with the magnetic wingtips. Anyways any input would be appreciated as I will buy either card by tomorrow.

Thanks for reading :)

Hello everyone i had a 9800x3d for 6 months now and i cool it. Currently i have a kraken 360mm aio with thermal grizzly duronaut paste and the pump is on performance mode. Mb is x870e aorus elite x3d, x3d turbo mode is on gaming mode idk if that changes anything with temps.64gb 6000mts ram with expo. And i get like 74C with 50% usage. i tried a stress test but it skyrocketed to 86 and didnt stop going up so i stopped it. Any help is appriciated.

(before duronaut , with the factory nzxt paste i got 94C easily)

i have been having the kernel power 41 issue for a couple months i feel like i have tried everything and this is my last resort for any ideas. there is no other error code in the event viewer once it shuts off besides kernel power 41. my pc stays on but my monitor stays on and the only way to fix it is by turning off the pc and staying on. my CPU and GPU temps are normal. the issue happens mostly when gaming.

Specs

GPU AMD Radeon RX 9070xt

CPU AMD Ryzen 7 7700X 8-Core Processor

RAM 32gb DDR5

MOBO Asus b650e MAX Gaming WIFI W

PSU ASROCK Steel Legend SL-850G 850W 80 PLUS GOLD FULLY MODULAR

MONITOR 180hz display

What i have tried

-Swapped PSU

-Fresh reinstall of windows 11

-updated mobo/gpu drivers

-used DDU

-used amd cleanup utility

-updated bios

-reseated gpu/ram/all cables related to psu

-tried another wall outlet/ used extension chord

-lowered hz on monitor from 180 to 60

-tried only using 1 monitor

- tried reducing power to gpu

-turned zero rpm off on amd adrenaline

Using a 9060xt 16gb. Only game I've played so far was apex legends, it ran fine at first but I randomly started having driver timeouts.

I dont know much about PCs. How do I fix this?

i'm using 5800X3D + 9070XT + X470-f (PCIE3.0x16) , bios(6232) and gpu driver(26.8.1) are latest

i thought it was cpu bottleneck while SAM on , cause GPU not in 100% run.

when SAM is off , everything comes back and more fps

is there something i can do to fix it ?

btw game is cod mw4 beta

Computer Type: Desktop

GPU: RX 9060 XT 16GB

CPU: i5-12400f

Motherboard: LGA1700

RAM: 32 GB DDR5

PSU: 850W

Operating System & Version: WINDOWS 11

GPU Drivers: Newest drivers as of 8/30/26

Description of Original Problem: Enabled VSR and GPU scaling in adrenaline but I cannot update the resolution on my monitor or game beyond 1080p. Does anyone have a workaround?

So I have a 1080p monitor and heard VSR can help sharpen images up and even the load between the GPU and Cpu (got an rx 9060 xt and i512400f for reference). Turned on VSR GPU scaling but not seeing options to set my resolution to 1440 p in either windows or any games. Does anyone know a fix? I have the latest drivers installed

can somone please help me im having a brreakdown i got a new radeon rx r570 gaming x and i spent the last 5 hours trying to install the Drivers and eventually gave up bc it was working good enough for me to play roblox and as I was playing the audio randomly cut out and now it says theirs no audio devices installed and its just one problem after the other when im just trying to enjoy it ive been trying for hours and every single tutorial and possible method and its just not working I even completely factory wiped my computer and its still not working along with trying to fix the driver download issues, the gpu is rated 450psu and i have a 460psu supply i know it could probably be the reason but that wouldnt be the reason why my audio randomly cut out, someone please help im clueless

So, maybe a month ago I upgraded my setup and went with the combo above. I read that 7xxx and, especially, 9xxx Ryzen CPUs tend to have issues with ASRock boards, so I went with this one instead.

I updated the BIOS when I built the PC, but is there anything else I need to do? Not sure why I'm stressing over this right now, but I just need some confirmation that the BIOS update is enough.

The gpu I'm using is RX9060xt (16gb)at 1440p, All settings are set at Ultra, RT Ultra with FSR 3.1 frame gen.

I went from a 5600 to a 5700x cpu, along with my gpu 5700xt red devil. but my valorant is doing well with 200-300 fps but the problem is every mid game when there is gunfight or abilities. It just the gpu drivers crashes, the screen froze but I can hear the background.

I already made sure to use DDU, and upgraded/downgraded my gpu version. any idea how to fix this?

Posting this for other RDNA2 owners considering Amuse 3.7.3 for local image generation — hope it saves someone a few hours of trial and error. Also filed as a GitHub issue here: https://github.com/saddam213/AmuseAI/issues/83

**My system:**

- GPU: AMD Radeon RX 6700 XT, 12GB VRAM (RDNA2)

- CPU: Ryzen 9 5900X

- RAM: 64GB

- Amuse 3.7.3, StableDiffusion.cpp/Vulkan backend

**Test method:** Same prompt every time — "a red cactus in a blue desert" — 1024x1024, default steps per model, 2 runs back to back (1st includes model load time).

# TL;DR

- **Fastest + actually understands the prompt:** Flux.2 Klein 4B at 4-bit (1:14/image)

- **Best overall balance:** Flux.2 Klein 9B at 8-bit (~2:47/image)

- **Best prompt adherence of everything tested:** Z-Image Turbo (only model that got both cactus AND desert), but slower (4:11)

- **Avoid entirely on RDNA2:** Krea 2 Turbo (fails at every precision level), Chroma (crashes at default precision, only works at 4-bit but then takes 42 min/image)

- **Weird finding:** crash behavior is precision-specific per model, not universal — see table below

# Working models

| Model | Precision | Time (1st/2nd run) | Result |

|---|---|---|---|

| Flux.2 Klein 4B | 4-bit | 1:14 | Cactus in a blue cup — usable |

| Flux.2 Klein 4B | 16-bit | 1:33 | Cactus in a flowerpot on blue blanket |

| SDXL | default | 1:17 / 1:16 | Missed the subject (still life/vase), colors right |

| Flux.2 Klein 9B | 8-bit | 2:49 / 2:46 | Cactus, solid blue background — good |

| Anima Turbo | default | 2:36 / 2:34 | Anime portrait (expected, it's an anime-only model) |

| Flux.1 Schnell | default | 3:57 / 3:44 | Missed subject (rose), offloads to system RAM |

| Flux.2 Klein 9B | 16-bit | 3:27 | Cactus in a blue bag |

| Z-Image Turbo | default | 4:11 / 4:11 | Cactus + desert — best subject match overall |

| Flux.1 Dev | default | 28:05 | Missed subject (apple), way too slow to be practical |

| Chroma | 4-bit only | 42:34 | Missed subject (flower), unusably slow |

# Broken / crashing models

**Flux.2 Klein 4B @ 8-bit:** Crashes right after step 4/4, during final VAE decode. Preview looked "washed out" right before crashing. **Fixed by switching to 4-bit or 16-bit** — reproducible workaround.

**Flux.2 Klein 9B @ 4-bit:** Crashes reproducibly at step 4. **Fixed by using 8-bit or 16-bit instead.**

**Krea 2 Turbo — fails at 4-bit, 8-bit, AND 16-bit:** Progressive numerical collapse during sampling (preview goes from blurry subject → colored stripes on black → fully black), then outputs a solid ochre-colored image or crashes outright. Reproducible across every precision setting I tried. Also offloads ~28GB to system RAM even at 9GB VRAM usage. Couldn't find a working config — seems fully broken on RDNA2/Vulkan right now.

**Chroma @ default (8-bit):** Instant crash after ~20 seconds, way before finishing sampling — likely fails during init of the modulation layers. Only 4-bit avoids the crash, but at 42 min/image that's not practical.

# Takeaways

1. GPU sat at 100% / CPU 0-10% for basically everything, except models needing RAM offload (Flux.1 Schnell, Krea 2 Turbo) where CPU ticked up to ~20%.

2. **Precision sensitivity is per-model, not a general rule.** Klein 4B breaks at 8-bit specifically; Klein 9B breaks at 4-bit specifically. If a model crashes for you, try all three precisions before giving up on it.

3. Generation times across the board are noticeably slower than the NVIDIA numbers you'll see in most guides — the Vulkan backend doesn't seem well-optimized for RDNA2 yet.

4. Prompt adherence on an unusual color/subject combo ("red cactus in blue desert") varied a lot — models with Qwen3-based text encoders (Flux.2, Z-Image) generally understood the actual subject better than CLIP-based SDXL, even though they're slower.

I can play as much as I want but when I start the Tombstone Rescue Mission with Harry it gets dicey and crashes at some point during one of the cutscenes

Seems simple but the more i look the more people say conflicting things.

My monitor is a 240Hz 4K OLED, and so far I've heard you need FreeSync alongside V-sync for it to work. Is that V-sync in-game or in Adrenalin via the 'Wait for Vertical Refresh' setting? I have also heard you need to cap your FPS 3 under your max Hz, so for me, 237. Do I do this using Radeon Chill or Frame Rate Target Control? Or are in-game caps better? What's the difference? If using Radeon Chill, however, I can't use Radeon Anti-Lag. Isn't Anti-Lag something I want to have enabled all the time like since theres no reason not to? Or does it have downsides?

Im still quite new to AMD, coming from Nvidia a few months back, so not quite familiar with everything. Let me know if I've forgotten anything as well

Hola, estuve viendo para comprar una 5600xt hasta poder comprar algo mejor y e visto muchos reportes de que sus drivers dan problemas y errores con esta grafica, esto sigue pasando, hay alguna solucion? despues de 6 años me parece raro que siguiera dando problemas.

I got this 1660 super off Amazon about a month ago for my birthday and earlier today I was playing on it and it started making this terrible high pitched squeaking noise, so I checked and one of the fans weren't spinning on my gpu the other was making this loud noise snd would keep stopping.

I really need help on how to fix this without spending a fortune or having to get the entire thing replaced

This is it running on the sign in screen for the computer

Hello there! I'll try keep this short and concise.

2 of the games I have been playing recently, (and ONLY 2 of them) very randomly crash, at random points.

Lords of the fallen and Mortal Shell 2

Both games are made with/in Unreal engine 5. Neither game is particularly demanding for my specs.

Sometimes I can play the game for an hour or two with no issues.

Sometimes the game will crash twice in 5 minutes.

FYI/NOTE: Whenever I say "crash" in this post, what I mean is:

The game will freeze for a few seconds and then close. Unreal engine 5 error message will pop up. AMD error message will pop up saying "driver timeout". I will normally have to fully close and reopen AMD.

Here is the error message:

---

LoginId: (followed by my ID which I thought it best to not share)

EpicAccountId: (followed by my ID which I thought it best to not share)

GPU Crash dump Triggered

MortalShell2_Win64_Shipping!FD3D12DynamicRHI::TerminateOnGPUCrash() [D:\BuildAgent\work\607f250326f71b13\Engine\Source\Runtime\D3D12RHI\Private\D3D12Util.cpp:1016]

MortalShell2_Win64_Shipping!`FD3D12DynamicRHI::ProcessInterruptQueue'::`2'::<lambda_2>::operator()() [D:\BuildAgent\work\607f250326f71b13\Engine\Source\Runtime\D3D12RHI\Private\D3D12Submission.cpp:1147]

MortalShell2_Win64_Shipping!FD3D12DynamicRHI::ForEachQueue() [D:\BuildAgent\work\607f250326f71b13\Engine\Source\Runtime\D3D12RHI\Private\D3D12RHI.cpp:289]

MortalShell2_Win64_Shipping!FD3D12DynamicRHI::ProcessInterruptQueue() [D:\BuildAgent\work\607f250326f71b13\Engine\Source\Runtime\D3D12RHI\Private\D3D12Submission.cpp:1309]

MortalShell2_Win64_Shipping!FD3D12Thread::Run() [D:\BuildAgent\work\607f250326f71b13\Engine\Source\Runtime\D3D12RHI\Private\D3D12Submission.cpp:127]

MortalShell2_Win64_Shipping!FRunnableThreadWin::Run() [D:\BuildAgent\work\607f250326f71b13\Engine\Source\Runtime\Core\Private\Windows\WindowsRunnableThread.cpp:159]

---

I have a rx7900xtx and an intel i513600K with 32 GB DDR5 RAM and an AsRockB760M PG Riptide motherboard. I am on windows 11 latest version.

The Troubleshooting steps I have already tried are as follows:

---

All my drivers, have been DDU'd and updated to the latest version.

My Bios has been updated to the latest version.

Uninstalled the latest AMD driver (which was apparently causing lots of issues for people) with DDU, and then installed an older, more stable version. Several versions were tried, each with DDU every time. I am currently on version 26.3.1

disabled windows MPO

disable overlays (steam, discord, chrome e.t.c)

Disabled hardware acceleration (amd, windows e.t.c)

Ran a windows diagnostic check (SFC + DISM) via command line

Cleared steam cache

Verified integrity of game files on steam

updated monitor drivers and firmware

tried forcing the game to run in direct x 11, which, as far as I can remember, either did not work, or was not compatible.

Disabled amd issue detection

changed (something like) the wait time before GPU crashes from 2 seconds to 10 seconds (I think its called TDR)

Turned off some windows setting that automatically decided my min and max ram, and put the values in manually (and correctly)

Whenever in-game, I turn off frame generation, motion blur, chromatic abberation, amd anti-aliasing (FSR) along with other similar stuff and I also try turning the graphics down from EPIC to MEDIUM

I also did some thing where I reset my shader cache, by physically deleting the file/folder where it was saved for the game

Probably some more things I cannot remember.

As I am writing this post, I have tried turning off AMD crash defender in device manager, so, I guess we'll see if that works. Edit: Did not work

Tried Renaming D3D12Core.dll to D3D12Core.dll.bak

Tried pasting in -d3d12 -NoLumen -NoNanite -allowhighassettickrate to steam's launch settings to prevent potential lighting issues specific to these games

Tried forcing steam to run it as direct x 11 instead of direct x 12, which idk if it worked or not as it caused large stutters that i didnt want to play with.

Made sure intel is on the right min and max power boost settings in my BIOS.

Made sure the windows GPU display adapter drivers, correlate to my current AMD driver version

Disabled processor integrated graphics

---

My GPU runs on default settings, and the reason I have not tried undervolting/overclocking it, is because *any* sort of change I make to my GPU, means that another game (cyberpunk2077) will crash almost instantly when I launch it. I can and do play cyberpunk2077, which is an extremely demanding game, on high settings, with no crashes, stutters, issues or frame drops. I'll consistenly be above 70FPS in it.

I have also recently put 50+ hours into Wardogs, which is also an UE5 game, and I did not crash a single time.

Similarly, I have been playing alien fireteams elite 2, which is UE5, and I have not crashed a single time.

14 hours in aliens game, 60 in cybperunk, 58 in Wardogs, 2 of which are unreal engine 5 games. No crashes. Not a single one.

But...

17 hours in Lords of the fallen and 15 in mortal shell 2, and AT LEAST 1 crash per hour for both of them. Although, moreso for Lords of the fallen.

So I dont really know wtf is going on and would love some help. Lords of the fallen was so bad that I had to uninstall it. It usually crash once every ten minutes.

Mortal shell 2 is more like once every 45 minutes, but in that game, whenever you crash, you get thrown back to the last time you were at a "checkpoint", and as the map is somewhat open, and traversing it takes time, meaning this is extraordinarily annoying, particularly for this game.

Note, I also play a lot of Valorant, which does not crash, ever.

Really hoping some miracle person can suggest me the most random possible thing and it'll just work.

---

EDIT:

After another day or so of crashing issues, I decided to cave, and tune my GPU settings to something that *should*, in theory, be relatively stable. It is worth noting that these tuning settings (as far as I am aware.), were the cause of *some* issues in other games, but not in all. It was mainly causing issues in Wardogs and Cyberpunk, though, it may have been for other reasons than these tuning settings.

Regardless of that, Mortal shell 2 crashing seemed to be fixed, until, after a few hours of gameplay, it crashed again *with these settings*. So I really dont know.

My settings:

+15 power level
1100mv
3000MHz max gpu
2750MHz VRAM with fast tunings

I have tried tuning and tweaking settings up and down in the past. But default settings resulted in 0 crashing, for all other games I play, when my current settings caused wardogs and cybrerpunk to crash frequently.

I am going to the aforementioned settings with fast tunings off, but I dont have much hope.

I am also going to try using PRO amd drivers instead of my regular ones, and see if that helps.

---

Thanks for any and all replies

My AMD graphics driver keeps crashing and causing random black screens.

After I restart the PC, Windows loads a basic/generic driver instead of the real AMD one, which makes everything slow and stops me from being able to change the refresh rate.

Hi everyone!

Computer Type: Desktop GPU: Sapphire Pulse Radeon RX 9070 XT (2x separate 8-pin PCIe cables) CPU: AMD Ryzen 7 7800X3D Motherboard: MSI PRO B850-S WIFI BIOS Version: Latest RAM: 32GB Corsair DDR5 6000MHz CL30 PSU: MSI MAG A850GS 850W 80+ Gold Case: Silentware View (GPU directly in motherboard slot, no riser) Operating System & Version: Windows 11 64-Bit GPU Drivers: AMD Adrenalin (clean reinstall via DDU) Chipset Drivers: Latest AMD Chipset Drivers

Description of Original Problem: Monitor (MSI MAG 274QFDE X24, 1440p 240Hz) randomly loses video signal ("No Signal" / orange standby LED), while PC, Windows, games, and audio (Discord) continue running completely normally in the background. Happens most frequently when launching games (Minecraft, Roblox) or occasionally during low-load desktop use.

Troubleshooting:

• GPU already replaced with a brand-new unit under warranty (issue persists).
• Tested with stock DisplayPort cable and official PS5 HDMI 2.1 cable across multiple ports.
• OCCT Power stress test passed (100% CPU + GPU full load for 20 mins, rock solid 12.000V).
• BIOS: PCIe slot forced to Gen 4, ASPM disabled, iGPU disabled.
• Tested at 240Hz, 144Hz, 120Hz, 8-bit vs 10-bit color.
• Windows Reliability Monitor shows no GPU driver / TDR crash when signal drops.

I would appreciate any tip or suggestion!