Graphics Cards

[superrob]

Hi all,

I recently wanted to try the AMD side of the fence as I haven't used AMD in years. So I got the opportunity to jump from my trusty RTX 3080 to a less then a year old 6900XT.

The microstutters were annoying still trying to get around those, the software was a bit of learning but I'm handling it now and the performance of the card was pretty impressive when you see the actual FPS jump from a 3080 to a 6900XT.

I have had so many BLACKSCREEN HARD LOCK CRASHES though.

The system I am trying to run this card on is a

Intel 10900k

4300mhz DDR4 32gb

Asus Maximus Formula z490

Antec High Current Gamer 1000w PSU

Asus ROG 24" 240hz 1080p screen

Windows 11

and all my drives are m.2 ssds

I've tried it all from DDU, to Windows tweaks, to Windows install, to gpu bios flashes, to stock gpu settings, to oc settings, to undervolting, to system OC, to no system OC etc

When the freeze occurs I need to hard power the system off or it stays in blackscreen forever.

Event viewer does not seem to be too discriptive with what has occured and all that Adrenalin says is the system has suffered a crash so all settings were adjusted to default or wording similar.

With the overlay metrics up all seems okay the gpu around 65 degrees in gaming due to my fan curve and junction temp around 78 degrees max, power limit maxed at 289w.

Any and all help is welcomed.

Most recently I'm usually running the record part of Adrenalin whilst gaming(I have it set as high as 100mb/s, 60fps, 1080p high etc to try get the best Youtube quality)

If I don't come right with this I'm afraid I will have to try acquire a 3090 instead :'(

[InvadersMustDie1131]

hello, whats your current driver?

[superrob]

Hi there,

I'm on the latest drivers from AMDs website. I have tried most fixes, at this point the only thing I haven't considered yet was changing the psu to something which is known to handle transient spikes best.

I have tried running the card stock as well and clocked down however it seems to have less crashes when overclocked a bit oddly.

Currently I'm even starting to wonder if it's not due to certain parts in the test system not playing well with the Adrenaline software for instance if there's some sort of conflict causing the system to enter Blackscreen and hardlock.

At first I thought it was me asking too much of the card with recording gameplay footage at the same time however the crashes are random, they never occur at the same point and they never occur at the same time frame. Some nights it feels crashless then the following night it'll crash once every hour.

All the software tells me is it was a hardware failure and that settings are reset.

In windows events it's pretty vague as well. The game logs also do not show a crash because technically it froze the system but the system did not reboot or bsod. I am even going to attempt to try a different displayport on the gpu just incase there is something strange like pin 20 occurring. 

Thank you for trying to assist I really appreciate it.

[Malaas]

I have the same card and the same problem.  Brand new rig and I’ve tried everything.  It’s a known problem for years and I’ve tried just about everything to fix it with zero success.  I’m done with AMD at this point and seriously dissatisfied.  For them to not fix the issue or something at all to address the issue tells me so much about this company and how they don’t stand behind their very expensive products. It’s back to Nvidia for me. 

[InvadersMustDie1131]

check the max frequency for your card on AMD web site then go to Adrenaline and set the max frequency - 100hz or 50 hz

For example:

Max frequency: 2450
Adrenaline max frequency set: 2350/2400

[Malaas]

I appreciate the advice but that doesn’t fix the problem.  This seems like it’s just a bad product

[JigglypuffNinjaSmash]

This frankly sounds like unstable memory tuning. 4x 8GB or 2x 16GB? Did you tighten the timings or just go faster? 32GB at 4300MT/s on a CPU weaker than Ryzen 5800+/Intel 12900k/13700k might be asking a lot. Go back to "stock" XMP and run some stress tests, then set your memory speeds/timings back and replicate.

What specific read-out do you get from Reliability Monitor (Start > "view reliability history") on those crashes?

[crspears]

Same card, had the same issue since owning it

SAM has something to do with it