General Discussions

[Wessmyre]

Powercolor 6900xt Ultra

   Crashes, Pc reboots, blinking, artifacts, crashing drivers, I can't go one session without one of these events plaguing my 6k rig. I thought it was my G9, bought the Sceptere 49".. same thing. Thought it was my PSU, new 1000 Thermotake platinum. Same thing. Thought it was a bad GPU, exchanged for a new one, nothing, same thing. Tried hundreds, HUNDREDS, of fixes over the internet... nothing, same problems.

  Reading this forum, literally hundreds of folks begging for an answer to solve these issues... nothing but silence. Not a single official response, nothing.  I have gone threw... 5? driver updates since I have had this piece of sht.... not one single fix, nothing. What was on the last one.... better game performance...? I DON'T NEED BETTER GAME PERFORMANCE.. the fking drivers are fundamentally broke. Fix this sht.

This is not a me problem, this is a you problem. Where do I send the bill for all this hassle and parts exchange? Who do I hold responsible for this mess?

[hrpuffnstuff]

Can you tell us whether or not your pcie plugs are each coming from different rails?

[Wessmyre]

Yes, yes, yes, Different rail, new power supply, prayers during a blood moon, dance with the devil at midnight and added two parts unicorn.

These drivers are ass. THAT is what needs fixed.  Not me attempting to fix it for them and hoping each trial will somehow magically polish this turd.

No disrespect, just done with the duct tape. Ready for a real fix. Enough with the over clock buttons and better game performance. **bleep** needs fixed, now.

[RPX100]

AMD hardware requires more knowledge and tinkering than nvidia/intel stuff.
Out of the box, nvidia/intel it is far easier to just "plug it in and forget about it".
Ranting about it in a user to user community forum will not fix your issues.

Try this: clean old driver with DDU, clean registry, sfc /scannow

And after that look up the max boost clock for your card.
AMD has set up the drivers to always boost as much as possible.
-> this means it will try to boost higher than manufacturer specs
-> you need to switch performance tuning to "manual" and enter your max boost clock

If you leave Radeon Software on "automatic" performance mode, then you will get crashes,
because it all is silicon lottery and boosting higher than spec = higher temps = more crashes.
(depends on silicon quality, yes: but this has been true and proven for RX 5000 series since 2019)

--
bad driver? well yes. Fixable? well, yes: with knowledge and learning about the cause.
Stop wasting your time with ranting - AMD is not listening. Start monitoring your card instead.
- Hot Spot Temp must stay below 90°C - GPU core to hot spot delta should stay within 15°C
- core / VRM / mem temps should stay below 75°C

[Wessmyre]

"clean old driver with DDU, clean registry, sfc /scannow"

Did it, several times. Actually worked..... for about 30 minutes.

" this means it will try to boost higher than manufacturer specs"

Did it, Made no difference both over and under.

"you need to switch performance tuning to "manual" and enter your max boost clock"

Did it, no difference.

"If you leave Radeon Software on "automatic" performance mode, then you will get crashes,"

Changed it many many times. Some are better than others, all broke in the end.

"because it all is silicon lottery and boosting higher than spec = higher temps = more crashes."

Again, broke out of the box, overclock, under clock, Wednesday while gaming, Thursday while staring at the desktop, Friday while eating breakfast, Saturday while reading the news, right now while typing this out, made no difference, broke.

"bad driver? well yes. Fixable? well, yes: with knowledge and learning about the cause."

The cause is the drivers, the fix is AMD getting off their as, acknowledging the problem and applying a fix. It is the ONLY fix.

"Stop wasting your time with ranting - AMD is not listening. Start monitoring your card instead."

That's all I have been doing because I sure as hel cant use the GPU for anything else.

"Hot Spot Temp must stay below 90°C - GPU core to hot spot delta should stay within 15°C"

Open water cooled tower. She never even gets warm.

" AMD is not listening"

  Right here, right here is the problem in the nut shell. this. FINALLY, a real answer. This isn't just a rant, this piece of sht is tearing up my rig. The driver is so broke Windows attempts to block the drivers for fear of hardware damage, STOCK settings. They sold me something that does not work for over two thousand dollars. There is no excuse, at this point and time this card isn't worth two hundred dollars, it's broke.

Lets face it, the ONLY reason this card is selling is because Nvidia is currently four thousand dollars, it's why I bought it, that and figured it would pair well with the AMD CPU, WRONG.

Here is yesterdays experience while using my PC...

Battlefield 4 (massive stutter, my 1500 dollar monitor starts blinking so bad I can't do anything to stop it, Pc crashes and reboots disabling the driver.)

Battlefield 2042 (Screen starts flickering so bad I can't really play. Did a restart.)

Mount and blade Bannerlord (Screen starts flickering so bad I can't really play. Did a restart.)

Mount and blade Bannerlord (massive stutter, unplug monitor and plug back in, driver crashes and reloads successfully locking up my game)

Surf Amazon for a new Nvidia card (Mouse artifacts show up, computer locks and restarts. AMD software pops error message on restart.)

Not that any settings in the AMD software will do sht, every lock resets ant AMD software to default.

And this is ONE day..Hel,  one HALF day.

[MADZyren]

EDIT: After re-reading your post, I noticed you mentioned having changed the GPU already. So: Fresh reinstall of Windows and all necessary drivers. If does not help, likely a hardware issue. To find culprit, try stuff mentioned below.

Commonly available hardware and drivers for them usually work. There are millions and millions of people using them and when most of them do not have issues, the problem is by far most likely in your machine, not in the design of hardware nor in drivers as if that was the case, everyone with that hardware/drivers would have issues and that is not the case here.

Have you tried running Prime95 to check the rest of your system runs stable?

Have you tried running Cinebench multiple times?

Are you running memory too fast or with too tight timings - easy check, disable XMS or equivalent and run it at 2133MHz.

Have you tried running 3DMark timespy to see if it runs ok?

While you could DDU the driver and try that, in your situation I would do a clean install starting from zero and after installing Windows, chipset drivers and GPU drivers, Steam(or similar) and one of the games you encoutered problems with, if you still have a problem, it is hardware related, aka. something is broken (this is why warranties exist) or installed incorrectly.

You mentioned watercooling. Did you install it yourself? Are you sure you did not drop any cooling liquid to GPU or motherboard? Did not forget any protective plastic on waterblock and you did rinse all the blocks before installing them? You did apply and spread enough non-conductive thermal paste (and did not use liquid metal!) You remembered to use thermal pads of exactly correct thicknes on all memory chips? You shifted your case from side to side with waterpump running so air got out to reservoir? You have connected all the required powercables to motherboard (the ones for CPU too?) and GPU?

Do you have problems will all heavy games or just some? If problems present themselves in pretty much all games (light ones excluded), your GPU is likely broken or damaged.

[Wessmyre]

With about quarter of a century of assembling my own computers and fixing other peoples PCs, I am inclined to say its not the drivers.

          Same, 100% driver related. Literally AMD pops up with driver errors within it's own self check. Wattman errors all over the map in AMD's own software reporting on stock settings out of the box. Windows event tossing out AMD GPU errors all over the map. Where do you get your information? "I am inclined to say its not the drivers." I can literally fill this page with AMD GPU errors from the last 7 days from both windows reporting and AMD's own software...  No offense, but you are wrong.

"Commonly available hardware and drivers for them usually work. "

This is exactly why I am here asking about driver fixes like hundreds of other people on this forum, and thousands on the net.

"Have you tried running Prime95 to check the rest of your system runs stable?"

Not prime95, but yes I have stress tested the entire PC. Stable. Interestingly enough the GPU did not fail during any of the handful of GPU related tests I preformed.

"Have you tried running Cinebench multiple times?"

Same results.

Fun fact, I can walk away from my rig right now and it will crash sooner or later. I can surf the net, play games, stare at the desktop. No rhyme or reason. Sooner or later the drivers fail. Sometimes I can Game on max for several hours no issue at all.. other times the driver will crash minutes in.

Are you running memory too fast or with too tight timings - easy check, disable XMS or equivalent and run it at 2133MHz

Memory is running at stock. I did under clock it after reading somewhere that anything above 2600 could cause issues. No change. Ran one stick, no change. Swapped one stick to #1 slot, no change. Same sht different configuration.

"Have you tried running 3DMark timespy to see if it runs ok?"

Again, the card runs fine. No overheat, no heat spikes, no voltage spikes, nothing to indicate an issue. While it's running it's running well within parameter. Even under load she runs like a champ. It's not the card, it's the drivers.  Literally can be watching tik tok and those drivers will melt. Windows will report it, AMD software will report it and toss up a "send report" and at times windows will block the drivers to safeguard the rig from damage, at times wattman settings will restore to default with an AMD error report.

"While you could DDU the driver and try that, in your situation I would do a clean install starting from zero"

Many times. This actually works... for a bit... 30 mins, hour, two hours.... then right back to driver crashes.

"after installing Windows"

This is a brand new build. This is the only card this rig has seen. Did a system restore on day one to have the ability to  "reset" rig to day one install. Tried it, again, worked for a bit, (hours) and right back to driver issues.

"If you still have a problem, it is hardware related, aka. something is broken"

Already did it. Sent the first one back to Amazon and got this one in. It worked great for a few hours, then right back to the same identical driver issues. Two cards, same identical issues, same in every retrospect. FYI I am not the only one mentioning this on Amazon reviews.

"You mentioned watercooling. Did you install it yourself? Are you sure you did not drop any cooling liquid to GPU or motherboard?"

Yes, I built the entire rig. It's hard piped, air pressure tested. was not filled until every component was in and tight. No leaks. Keep in mind I RGA'd the GPU for replacement so it's not the same one.

"Did not forget any protective plastic on waterblock and you did rinse all the blocks before installing them?"

Of course, but this would have shown as a heat problem anyway which I do not have. All sources are heat monitored, this rig runs cold.

"you did apply and spread enough non-conductive thermal paste (and did not use liquid metal!) "

Yes, top of the line paste. Course, again this is all heat related. I have no heat issues at all. I also have no CPU events, only GPU. Also did not mention, I have top of the line DP cables installed. Bought 3 tried two and the one that came with both monitors. Same GPU issues as stated.

"You remembered to use thermal pads of exactly correct thicknes on all memory chips?"

Heat sinks Built on RAM. Ram is air cooled (separate fan) in an open tower, its not enclosed. No RAM events in windows. I don't have more RAM, but this is what I did... Under clocked Ram. Took out all ram and re-seated slot one only. Then changed ram stick out in slot one with a different stick, same set. All ram booted no errors. Tested ram, all four sticks, ram test passed.

"You shifted your case from side to side with water pump running"

Yes, this is a massive pump, it's very big. Took several hours + 24 with dummy pin in to bleed the system. Again heat related problem. I don't have that.

"You have connected all the required powercables to motherboard (the ones for CPU too?) and GPU?"

Yes, everything runs fine without any windows events, with the exception of the GPU. All buttons, USB, reset, everything works.

"Do you have problems will all heavy games or just some?"

Some facts...

All games that I have played so far, it's an extensive list, have the exact same issues. stuttering, PC restart , artifacts, crashes, lockups. All of them have shared this at one point or another. All of them have had more than one of these issues.

This also applies to internet surfing, streaming, idle PC, You Tube, Same set of issues. I can literally walk away and the drivers will fail sooner or later.

Another fact. EVERY TIME  i boot the failure will happen within the first 5-30 minutes.  One of those issues WILL happen no matter what I am doing. After that first initial restart it becomes semi stable and can take up to an hour before it happens again "sometimes sooner" AMD, and Windows Event both see it as a driver related issue, almost every time. Most of the time it comes back up with the AMD report tool window and watmann reset notifications. Yes, I already tried the "disable fast boot" fix, same outcome.

[RPX100]

Did you try to disable all power saving features / deep sleep modes via MorePowerTool?

I had random and rare green screen crashes while running older / CPU bound games.
Disabled all the "DS_" (deep sleep) features for my card and green screen crashes are fixed now.
On top of that it also fixed the stuttering and occasional screen flickers are gone now.
edit: make sure your GPU memory clock is stable - if it is jumping up/down a lot = crash

Downside of course: more power draw while idle - as in: 11-12W before; 14-38W idle now *sad face*

[Wessmyre]

"Did you try to disable all power saving features / deep sleep modes via MorePowerTool?"

Did it within windows: Current settings:

Screen: Never

Sleep: Never

Under advanced:

Ultimate performance

Display never

Sleep Never

"I had random and rare green screen"

Ya, I read about that, for the record I DO NOT have the green screen issue.

"Disabled all the "DS_" (deep sleep) features "

Yes, all sleep options are disabled. Also tried disabling Freesync in both monitor and AMD software. No dice, same issues.

[RPX100]

The settings you are referring to are Windows settings. They are not helpful here.
I am talking about deep sleep features of the GPU (driver controlled).

MorePowerTool (link) is a community made tool to change the behavior within the driver itself.
Read the article to understand what it does.

You can disable all the power saving features with this tool.
This is not possible via Windows / Radeon Software. You need MPT for this.

[MADZyren]

"Heat sinks Built on RAM."

Yes, but VRAM?

"Same, 100% driver related. Literally AMD pops up with driver errors within it's own self check. Wattman errors all over the map in AMD's own software reporting on stock settings out of the box. Windows event tossing out AMD GPU errors all over the map."

Error messages simply mean there is something wrong, not that it is a driver issue - a bit like if no water comes out of the faucet, it is not necessarily a faucet issue.

If problems occur only on some games, then it can be a driver issue, but I got the impression that it is not just some games. I once had a broken CPU which was a bit like that. Sometimes it worked fine for some time. Sometimes if crashed fast. Finally, I had to reboot computer time and time again and reinstall Windows to get it to run even for a short time. When installing Windows, it usually worked flawlessly and troubles began only after restarting Windows after installation. Perhaps operating system uses some parts of CPU when you actually run operating system that are not used when installing.. maybe some register, cache or something.

In your case, I would install windows and required drivers, no wattman, nothing extra, reset bios (as, if I recall right, wattman overrides bios settings and even if you unistall it, does not necessarily default things), disable XMP. If you have multiple hard drives, unplug the rest except system drive. Unplug power and reset switches from motherboard and all RGB headers, front panel USB too. 

EDIT: A switch might seem to work, but can still cause issues. I think some youtuber who fixes computers also had an example of this.

EDIT2: Check BIOS to confirm that your bus speed is 100.00 and nothing higher as this can cause very weird issues! I would probably set an all-core clockspeed of say 3700MHz to CPU@1.1V (assuming a Ryzen CPU). You could also try Linux and not install any graphics drivers. 

EDIT3: When you got a new PSU, did you change cables too or only PSU and used old cables?

[BubbaNoy]

Bad Drivers!!

I'm still fighting with my 5700xt.

Driver freezes on mine randomly while web browsing or just updating my programs. Low impact.

I've already sent it for RMA just to get another 5700 xt that does the same thing, so that leads to one issue. BAD DRIVERS!!

While it was on RMA, my old GTX 960 was rock solid, just that the game performance was  **bleep**. But it was ROCK Solid, no crashes issues at all.

[MADZyren]

It's just that I have a Powercolor Red Dragon 5700XT and I have played many games and use this machine many hours a day, but do not really have noticeable issues. Sometimes Radeon software has reset my OC settings but I have not noticed any crashing, slow-downs or other issues. Recently I've played FC6, Horizon Zero Dawn before that, Doom Eternal before that.

You are not the first one to have issues with 5700XT though. I would so much like to find one of the poorly working GPUs or machines just to try to find the problem as I find it very interesting.

[RPX100]

@MADZyren (you have private messages disabled - posting my message here instead)

my personal journey with RX 5700 XT:
https://community.amd.com/t5/graphics/asrock-rx-5700-xt-challenger-pro-triple-fan/m-p/424096#M67706 

more recent update of the whole story:
https://community.amd.com/t5/drivers-software/black-screen-when-using-amd-freesync-in-new-world/m-p/... 

Been building my own PCs for 20 years now (including for friends and family) - never encountered so many issues before.
Was actually going to give this PC to family after 3 or 4 years (as we usually do) - but not sure if I want to give them these problems.

Ask me anything, if you are that interested in the whole issue. ^^
Also check my profile for posts with even more details on Radeon Software issues.

[MADZyren]

Automatic overclocking settings of Radeon software (if you allow it to determine overclock/undervolt settings aso.) are all over the place. In my case the GPU would be quite unstable with those.

My current settings are Frequency/Voltage: 800/750, 1440/838, 2081/1169, powerlimit +5%, memory at 1800 MHz (which is already pushing it - memory chips on this Powercolor simply are not the greatest ones). SAM is enabled, Radeon Enhanced sync is disabled (though I had no issues, but have heard it caused problems to some people with earlier drivers).

Screen going intermittantly black sound a bit like what happens if I OC VRAM too much. Set it to 1750 if you already have not tried it. I have wondered, could freesync implementation in some monitors cause these issues. With GSync you have Nvidia hardware inside monitor. With AMD... it might be that every company can produce the functionality of Freesync as (cheap) as they wish. (I have Acer 27" 1440p freesync). 

I encountered black screen on one game a couple of times (Borderlands 3). The second time it happened I desided to just wait and after a looong time, picture came back and it was not a driver crash (I checked). I think it was doing something with shaders or whatever Far Cry 6 does (Warming up VSO cache). This was when starting the game. During gaming, I don't think I've ever had an issue with this GPU.

Not so long a go I watercooled this to make it quieter. Now it is great because you can do anything and noise levels just stay the same.

I just don't get it where the problems some people have with these GPUs could come from. I wonder if it is specific makers or models only or have some faulty GPU chips passed QA-testing somehow and cause the issues some users have.

[Wessmyre · Accepted solution]

Ok, I need to eat some crow. I'm humbling myself here just in case it can help someone else.

      I bought some very high end display port cables, 8x's to be exact, very expensive, not junk. Two came in the box. Needless to say the first time i ran testing I changed out the cable with the backup cable from that box. Problem persisted.  

     When Initially wrote this I was absolutely positive this was a driver/GPU issue. Keep in mind I swapped out three, THREE monitors in testing. Two G9's and a  Scepter 49" all with the same exact problems. Changed cables with the same outcome. I was literally at the end of my rope. The day I wrote this, being fed up, I jumped on Amazon and purchased an Nivida RTX Ti which arrived yesterday. I dropped it in, cranked it up and.......   Same exact problems, not a single solved issue.  

     I ripped my rig down and started chasing wires, one at a time. updated anything I could and reset both ram and CPU, cranked it up and...... Same problems. Please keep in mind I didn't have a "video" problem, at times my entire rig was being brought to it's knees. GPU errors everywhere, almost immediately on startup. 

   As an afterthought "and desperation" I grabbed an old HDMI cable and dropped in and BAM! EVERY. SINGLE. PROBLEM. disappeared. I reinstalled the Powercolor 6900XT ultimate and she was smooth as silk. I had my wife run to Bestbuy last night and grab a cheap DP cable and dropped it in, and she is officially fixed, my PC, not my wife, she's still broke..

(I thoroughly tested my PC)

Three things...

1- Check the cables if you are having issues, don't over think it, just do it. 

2-My apologies for being short the other day to folks that were trying to help. I was having a full blown man fit. I can accept this.

3- To the fella that reads the AMD error reports, my apologies for the profanity laced tirade.

Lesson learned.

[MADZyren]

Great to hear you found the problem. Often it is the last thing anyone would have thought of.