Graphics Cards

[jhautrey]

I have had a consistent black/green screen crashing issue for about a year now which seems to come and go, but these last 3 weeks it has been nonstop. I've tried so many fixes for the issue, none have had any effect, such as DDU, all types of graphics tuning, BIOS updates, drivers, undervolting/underclocking/overclocking ETC and NONE of them have had any effect on the crashing.

My situation seems a little bit different from those who normally have the black screening issue, as ANY type of movement inflicted on my PC usually guarantees a crash. All cable connections are secure, though I have noticed once in a while that adjusting the HDMI cord into the card sometimes causes a crash too. A black screen crash typically ends up in a restart which then causes a loop of random crashes from opening up different programs (google, discord) which come at random, and reports as a driver timeout whenever I am lucky enough for the display to come back after black screening. The loop of crashing is usually fixed by waiting a whole day or night until I get on at a later time, or reseating the graphics card (usually my solution).

I have laid the PC onto its side which seems to make the random crashes less frequent, though they still happen.

This is starting to piss me off as I have also recently bought MWII and I am not able to play a good 5 minutes of it, even on the menu screen, without it black screening and needing a restart. MWII seems like the best trigger for the crashing (besides moving the pc) though other large programs have also caused it (Sea of Thieves, Bloons TD).

I really don't want to have to go get the PC checked out but I will need to soon if I cannot fix this.

[Astig]

Ever since I installed the newest version of the drivers, I've been crashing every single game now. I can't downgrade cause there's no option to choose a specific version. 22.11.1 is the newest and current version I have and I think that's whats causing the crashing/freezing every game

[jhautrey]

You can uninstall your adrenalin drivers and then select an earlier version on the download page. Unfortunately this does not solve my issue, but it may solve yours.

[Volkdude85]

what wattage is your PSU?  

[Volkdude85]

https://www.guru3d.com/articles-pages/gigabyte-radeon-rx-5600-xt-gaming-oc-review,8.html

your 177 Watts according to that article.

That's a spicy lightbulb.  14.75 amps on your 12 volt rail after it has been rectified and not peek.

Id check you have the power. 

Good luck.

[jhautrey]

I have an Asrock AMD Radeon RX 5600 XT Challenger that came in the prebuilt that I am using. I'm pretty sure it gets enough power as I have not switched anything out of the PC since I have gotten it.

[DimkaTsv]

All cable connections are secure, though I have noticed once in a while that adjusting the HDMI cord into the card sometimes causes a crash too.

Are you really sure this cable is of good enough quality? Bad cables can cause nasty stuff with image output on AMD cards. 
Flickering Issues With AMD GPUs? We've Got You Covered! Driver Settings And Choosing Proper Cables -...

[jhautrey]

Just tried removing both HDMI cables from my 2 monitors and used a single DisplayPort cable to 1 of them. Was able to play MWII for about 5 minutes until it black screened again. I'm not sure of the specifics to these cables but the DisplayPort cable also supported my 144hz.

[DimkaTsv]

Hmm... after it black screened... Did MWII crash with any error code? 
If not then try either globally disable HW Acceleration and Vertical Synchronisation
Or you can try to disable MPO... As you run 2 monitors, disabling MPO can help with a good chance.

To disable MPO = regedit add this key
[HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\Dwm]
"OverlayTestMode"=dword:00000005

To enable MPO back if you ever would want that, just remove that same key you added. 

[jhautrey]

Disabled MPO, still got a crash. However, this time I got double black screens upon crashing. Typically when others experience this crashing issue, one monitor goes black and one goes green (only when two monitors are active). This probably doesn't lead to anything though.

Also, I've been noticing that sometimes these crashes are triggered by small events that happen on the PC. For example, last crash I specifically was playing MWII and got a message on discord within a second before the crash. Another time it crashed as someone joined a call that I was in (again, while playing MWII). My theory is it has to be driver related and that they really dont like other processes interfering with MWII. Though I have had the same crashes while simply watching a screen share, so I'm still very lost.

[jhautrey]

Nevermind, I just crashed again within 2 minutes of opening MWII w/o anything else running in the background.

[DimkaTsv]

Did you alt-tab?
MW2 is really sensitive to one current problem. For me solution was to disable GLOBAL VSync (Force off). As well as disable one in game. I hadn't crashed with alt-tabbing since yet.

But fact that you disabled MPO and still crashed. . . Hmm. Maybe it is just vsync that broken currently?

[jhautrey]

Actually, the first time I alt-tabbed a few times and noticed a gray screen for a second while switching tabs, and then later I crashed. I saw the issue with alt-tabbing but it may not have been the cause of the crash.

However, the second time I ran MW2 with MPO disabled AND no other programs open, I did not alt-tab at all and still ended up crashing.

Just checked Vsync and I had the setting on "Off, unless application specifies." I'll try completely disabling it and playing right now.

[DimkaTsv]

However, the second time I ran MW2 with MPO disabled AND no other programs open, I did not alt-tab at all and still ended up crashing.

MW2 should report error after crashing in such way. If it is 0x00001337 (0) - then it is one that happened after alt-tab for me and was solved with vsync disabling.

0x887a0005 on other side was happening if i overclocked my GPU too much, with heavier OC leading to faster crash.

[jhautrey]

MW2 should report error after crashing in such way.

I did not get an error code, it was a complete black screen crash which required a restart.

0x887a0005 on other side was happening if i overclocked my GPU too much, with heavier OC leading to faster crash.

After completely disabling Vsync, I went on to play a few minutes and as I was changing some game quality settings I went through a blinking black screen for about 20 seconds until I finally got a COD error code (I have not gotten one of these yet, maybe disabling Vsync saved my PC from a complete crash?).
The error code was a 0x887A0005 like you said. My AMD tuning is all at default right now, and I have not overclocked anything. Anything you suggest I do?

[DimkaTsv]

The error code was a 0x887A0005 like you said. My AMD tuning is all at default right now, and I have not overclocked anything. Anything you suggest I do?

Hmm... 0x887A005 from what i understand means DXGI_DEVICE_HUNG error analog. I can suggest try downclocking 100-200 mHz?
Go in Wattman and check custom OC on GPU.
Check "GPU Tuning"
Check "Advanced control"
Then you have several choises. Any driver crash (and frequency related crash is ALWAYS driver crash) will reset OC profile to default values.

1) Try to set minimal frequency to current max frequency -100, so if you have max frequency = 2342 , then you set minimum frequency to 2342-100 = 2242. This one resolves transient response related issues.

2) Reduce max frequency by 100-200 mHz. This will resolve GPU frequency instability related issues. Performance hit will be minimal, but you will have more stability. But!. Remember default max frequency value you have. This specific number have different behaviour than any other you custom set (at least for my RX 6000 series). Basically when i have 2654 or 2659 for default (it can actually randomly vary for some reason, but this specific number you are seeing by default, actually will clock higher by about 25-30 mHz. Any other number besides default will get approximately -50 mHz penalty hit.
So if i have default 2659, i will clock up to 2685 approximately. But if i set 2658, i will only clock around 2610. Currently my default is 2654, and if i set 2659 (which is not my CURRENT default) i will get penalty hit and only get 2610 as result
Basically....
Current default number ---> default +30 mHz
Anything else ---> Set number -50 mHz.
Just take it in mind.
If you set back default frequency as it was before, you will get default behaviour again so don't worry.

3) Do both min frequency and max frequency tweak at the same time.

[jhautrey]

Tried reducing minimum frequency by 100, 0x887A005 error.

Tried only reducing max frequency by 100, full computer crash & restart.

Tried combo of both, 0x887A005 error.

Whenever there's a COD error, theres also a AMD driver timeout notification.

Don't know where to go from here. Gonna try 200MHz decrease.

[DimkaTsv_alt]

Hmmm...
As additional steps you can try disabling or uninstalling the software that controls the lighting effects. eg: ASUS AURA lighting effect, Gigabyte RGB FUSION, MSI mystic light... -Turn off Overlay on NVIDIA, Discord, AMD... -Disable Razer Cortex's booster and all same app.

[jhautrey_alt]

I don't have any software that controls my lighting effects.

Turned off AMD Adrenalin/Discord overlay, though it did not do much considering I was crashing without any of these programs running.

[u7QO8]

it originally is 5, u dont need to add a regedit and it is there already.

[DimkaTsv_alt]

If you have this key at all, this means that MPO is disabled for you
To enable MPO this key must be deleted.

And yes, by default, there is no such key.

[jhautrey_alt]

Bump

[jhautrey_alt]

Bump

[jhautrey]

Recently noticed that I get "a fatal hardware error has occurred" in event viewer. Event ID 18.

A fatal hardware error has occurred.

Reported by component: Processor Core
Error Source: Machine Check Exception
Error Type: Cache Hierarchy Error
Processor APIC ID: 10

The details view of this entry contains further information.

Everything online related to this error says to do things I have already tried, such as updating BIOS, drivers, overclocking, etc. I'm starting to think it's just something on my board that is defective.

[DimkaTsv]

WHEA error 18 is related to too high FCLK that system is unable to hold up stable
Processor APIC ID shows what core had thrown an error before failing compute.

You can try stresstest memory and look for WHEA errors line at very bottom of HWINFO
Or if it is somehow CPU voltage issue, then you should try something like prime95 or y-cruncher or CoreCycler. They load system heavily. But they are also very sensitive.

[jhautrey]

I started a 3 minute stresstest on my GPU with AMD software and at around 1:30 elapsed I experienced a black screen crash. Now I know it is definitely caused by stress to the system, but what would I do to find a fix? I do not know how to track it using HWINFO or what I would do with that information.

[DimkaTsv]

Tbh HWINFO check is only for ensuring FCLK stability. And as you are at point when even AMD software stress test fails (it is quite mild on sensitivity from my experience), there is something serious going on.
Are your CPU still on warranty?
Basically you would want to check potentially faulty components on different system to insure it is specifically what causes issue. Then RMA it as faulty chip. If possible.