New gaming PC (RX6800xt) video crashing

Forums › Desktop computing › New gaming PC (RX6800xt) video crashing

floydie

474 posts

Ultimate Geek
+1 received by user: 49

#315584 27-Jul-2024 21:17

Hi everyone.
My son just built his first gaming PC.
He keeps getting the display black screening like the display port has gone out.
System is....
MSI pro b760 wifi
Intel i5 12400
Corsair vengeance lpx 32gig
Crucial p3plus 2tb M2
Graphic card -XFX speedster swift 319 Radeon Rx 6800 16gb
Corsair RM750e psu

I've seen lots of different "fixes" from undervolting to firmware rollbacks and MPO turn off, and finally DDU (whatever that is)

And finally I'm not 100% on the PSU connections to the GPU...its modular and I'm assuming it 2 rails but the sockets aren't labelled as such so not sure I I should be using the two top sockets or side to side or what to spread the load.

Any help on getting this fkn expensive paper weight working?google has hundreds of 6800 crashing issues but no real answers.

Batman

Mad Scientist
30014 posts

Uber Geek
+1 received by user: 6217

Trusted

Lifetime subscriber

#3265301 27-Jul-2024 21:42

it's a AMD issue

find a firmware version that's the most stable

or buy Nvidia

shrub

790 posts

Ultimate Geek
+1 received by user: 272

ID Verified

#3265339 27-Jul-2024 23:02

Not its not an AMD thing. More than likely a driver conflict. Windows updates likes to install old drivers over new ones on fresh installs.

Definitely try DDU first. Very easy to follow guide Here

https://www.wagnardsoft.com/content/Download-Display-Driver-Uninstaller-DDU-18079

Also grab the latest AMD drivers Adrenalin 24.7.1 (WHQL Recommended) and have ready to go after DDU.

https://www.amd.com/en/support/download/drivers.html

Power supply does not matter which ports on the PSU as long as you run 2 separate cables. That one has 1x 12v rail.

The only other thing to check is temperatures when stress testing monitor with HWMonitor. On my 6800xt the hot spot is around 85-90c and GPU max is about 75c. Also consumes around 300w.

Jase2985

13735 posts

Uber Geek
+1 received by user: 6216

ID Verified

Lifetime subscriber

#3265488 28-Jul-2024 10:48

Batman:

it's a AMD issue

find a firmware version that's the most stable

or buy Nvidia

where is your evidence to back this?

you have overlooked even the most basic of faultfinding

Qazzy03

546 posts

Ultimate Geek
+1 received by user: 492

#3265521 28-Jul-2024 12:44

It sounds a bit random but have you tried another different DisplayPort/HDMI cable and or another output port in the GPU?

I have had a couple of issues in the past, where the problem was the cable/monitor issue connection issue.

*Edit*

I have owned a 6600XT and currently own a 7900XTX.
AMD can have driver issues but from what I have encountered it is usually crash to desktop, rather than lose connection to display.

SpartanVXL

1498 posts

Uber Geek
+1 received by user: 666

#3265526 28-Jul-2024 13:19

I’m assuming you’ve tried the undervolting, MPO and driver rollback? Drivers as far back as 23.11.1?

If you’ve singled out the GPU then send it back as faulty.

floydie

474 posts

Ultimate Geek
+1 received by user: 49

#3265566 28-Jul-2024 15:17

Thanks for the suggestions.
Driver is 24.7.1 downloaded from amd site.
I read that amd software can overclock the GPU even at default settings...
Default adrenaline Max speed was set to 2250Mhz
And by looking online the specs says clock frequency should be 1815mhz and "boost clock speed " around 2000...
So I've set max clock speed to 1850 and had the offending game open for about 3 hours with no crashing. Clock speed seems to be hovering around 1788Mhz...
Not sure if I should set max clock speed up to the "boost speed" of 2000mhz?
Some sites say 2105?

Mighty Ape

Shop now at Mighty Ape (affiliate link).

K8Toledo

1018 posts

Uber Geek
+1 received by user: 311

#3265615 28-Jul-2024 19:46

floydie: Hi everyone.
My son just built his first gaming PC.
He keeps getting the display black screening like the display port has gone out.
System is....
MSI pro b760 wifi
Intel i5 12400
Corsair vengeance lpx 32gig
Crucial p3plus 2tb M2
Graphic card -XFX speedster swift 319 Radeon Rx 6800 16gb
Corsair RM750e psu

I've seen lots of different "fixes" from undervolting to firmware rollbacks and MPO turn off, and finally DDU (whatever that is)

And finally I'm not 100% on the PSU connections to the GPU...its modular and I'm assuming it 2 rails but the sockets aren't labelled as such so not sure I I should be using the two top sockets or side to side or what to spread the load.

Any help on getting this fkn expensive paper weight working?google has hundreds of 6800 crashing issues but no real answers.

I would first check Windows Event Viewer/Administrative Events for any logs.

When does this occur?

SpartanVXL

1498 posts

Uber Geek
+1 received by user: 666

#3265772 29-Jul-2024 10:06

Boost speeds are opportunistic, it will go faster if it less hot and has power headroom. The speeds listed as default don’t really matter.

If you can spot the temperatures while it’s running then that will let you know if it’s hitting a wall anywhere. People recommend undervolting to avoid it hitting the hot zone.

If it’s running stable at slower clocks then it is either not getting enough voltage, or it’s overheating. Given how the top recommendation is to lower volts, stock voltage should be fine for 2.2GHz.

K8Toledo

1018 posts

Uber Geek
+1 received by user: 311

#3265786 29-Jul-2024 10:45

Historically certain AMD GPU's have been prone to TDR errors which manifest as black screens (often mentioned in the driver release notes).

Bumping up the voltage could sometimes fix this issue.

As for the OP's case it's hard to say without more information.

SpartanVXL

1498 posts

Uber Geek
+1 received by user: 666

#3266530 31-Jul-2024 11:04

I mention temperatures first and foremost because recent AMD gpu’s have had issues with their vapour chamber coolers and/or bad contact causing high tjunction readings.

If temperatures readings are fine in hwinfo then raising the voltage a smidge is good to test. Otherwise it would be knocking the core clock down a little by 50-100MHz, @floydie give that a try. -200 to -400MHz is a bit too much performance loss for a brand new card.