Graphics Cards

[sketterman]

Hello All,

To preface - please forgive the lack of brevity here...I tend to explain in as much detail as possible; you never know which detail might be the important one.  That and I've been screwing around with this PC all day today, and I'm kind of on my last nerve lol.

With that out of the way, PC info:

OS:  Windows 10, 22H2

MOBO:  ASRock Z690 Phantom Gaming 4, all stock configuration

CPU:  Intel 13600K, all stock configuration

GPU:  ASRock Phantom Gaming Radeon 6950XT, all stock configuration

PSU:  Thermaltake Toughpower GF3 1000W

RAM:  Kingston Fury Renegade 32GB (2 x 16), 4000Mhz DDR4

HDD:  Various SSD's; all games are on a Samsung 860 EVO; Windows is on a 980 PRO

The above machine was built by me back in April of last year.  That part is relevant only because all of the issues below started happening about a two weeks ago.

About 2 weeks ago, I got the itch to do another playthrough of Doom Eternal.  So I installed it, and when I loaded it, I noticed the following:

- While the framerate itself was buttery smooth, the movement was 'jittery'.  In a sense that it seemed as if the character was walking, then running, then walking, running, back and forth. 

- The FPS counter (via Steam) displayed a crazy metric; frame rate was jumping between something in the 100's, to something in the 300's, and even into the 400's.  It was going back and forth so fast that it was nearly impossible to see the numbers.  Constantly doing this.

The game did not display this behavior in the past.  I updated Adrenaline/GPU drivers which didn't help.  For that particular game, enabling the in-game V-Sync fixed the issue.  Frame rate was steady at 165, and the weird movement problem went away.  Did an entire playthrough with no issues.

Fast forward to yesterday.  For whatever reason, I wanted to play Crysis 3, which was available on PC through Game Pass.  So I installed it and launched it, and the performance was horrendous.  Like Eternal, the game ran fine the last time I had played it.  Unlike Eternal, though, the frame rate would not go above 63 FPS.  If I maxed out the settings at 2k reso?  63 FPS max.  If I lowered the settings to 'Low' and lowered the reso to 1080p?  63 FPS max.  GPU usage, via Adrenaline, never really went above ~40%, and the "micro-stutter' metric hung out between 20%-40% the entire time.

So I started messing around with Adrenaline, making sure everything was updated, and eventually disabling pretty much everything.  Nothing fixed the performance issue.

Thinking this could be some weird 'bug' with the older game engine and my hardware,  I hopped on Steam and found the entire Crysis 'Remastered' bundle for like $25.  So I bought it today, and first downloaded the 1st Crysis.  Booted it up, and it does the same thing Doom Eternal does.  Movement is 'jittery', again like I'm moving at one speed, then slowing down, then speeding up, slowing down.  Non stop.  No graphical settings, either in-game, or via Adrenaline, fix this issue.

Next, I installed Crysis 2 Remastered.  When I booted that game up, the 'jittery' issue was so bad it was unplayable.  I didn't even bother troubleshooting the game, I turned to the Google machine.

Found this link:  https://community.amd.com/t5/graphics-cards/6950xt-stuttering-in-games/td-p/529469

Followed the guidance that is marked as "Solved!", provided by user Justifier, nearly to a T.  This involves disabling any and all 'sync' options in monitors, windows, adrenaline, and in-game, uninstalling unused programs, registry edits, all kinds of stuff.  None of that fixed the issue.  So I scrolled down that link a bit and found another user who offered a link to a Youtube video with some other steps you can take.  Followed those steps, and still no resolution.

At this point, I figured the issue has to be something related to the GPU drivers, as this exact system was fine up until a few weeks ago.  So I decided to go through my Steam library, and play each game that I have installed, just for a minute or two, to see which games act up.  Some of my games still play fine, but the ones that dont?

-Borderlands 3:  Jittery issue isn't present when I launch the game, but as I play, it gets worse

-Crysis 1 Remastered

-Crysis 2 Remastered

-Doom Eternal (but is fixed with V-Sync enabled)

Here is where things take a turn for the worst.  As I'm loading these games, one at a time, I eventually make my way down to Metro Exodus Enhanced Edition.  Within 20 seconds of loading up a save file, my PC shuts off.  No warning, no alerts, no indicators, just shuts off.   My monitors and audio interface did not lose power.  It wasn't a power loss, or a fuse issue, just the PC.  So I power back up, load Metro again, and again, PC shuts off.

At this point, I'm assuming the changes I made to my PC via that link above caused this, so I go back and revert everything to how it was (minus reinstalling old programs).  Then I load up Metro a 3rd time, and my PC shuts off again.

So now I'm looking into this, most results I find are pointing toward GPU temps.  So I install Hardware Monitor, open it up to display both CPU and GPU temps, and start loading my games AGAIN, one by one, and annotating in a freaking spreadsheet both GPU core and GPU Hotspot temps, and it was alarming:

At idle, my GPU hovers in the low 50 C, and the hotspot is in the upper 50 / lower 60 C.  That's how it is right now as I type this.  No game has been running for about 30 minutes.

Additionally:

Borderlands 3:  GPU Core reached 102C / Hotspot reached 116 C (less than 1 minute of in-game time)

Borderlands GOTY Enhanced:  nearly the same; 101 C and 111 C respectively.

When I loaded Crysis Remastered, the last temp I saw with my eyes was the GPU Hotspot was at 115 C.  My PC shut itself down shortly after I saw that.  Once that happened, I decided to come here.

- I opened up my PC case.  It's slightly dusty inside, but overall it's clean. 

- Case filters are clean

-All case fans are running with no issues; CPU fan and GPU fans are all clean and running fine

-GPU is seated properly, no power cables are loose or damaged.  They are all new as of last year, and I haven't even needed to open the case up until today.  

I don't really know how to even recap all of that.  Some games have jittery performance, some have poor performance, others run perfectly.  My GPU temps seem extremely high based on the searches I've done.

Other things worth noting.

-My previous GPU was an NVIDIA GTX 1080.  I replaced it only because I wanted better performance, as it still worked fine.

-Right after installing the 6950, I was in the middle of a Fallout 4 playthrough.  The performance with the 6950 was atrocious.  I don't even recall what I had to do to get it to play right, but this GPU did not play that game well out of the box.

-I own Doom 3:  BFG Edition.  This game will not even launch.  No audio, no video, no alerts, no warning, no errors.  I click to launch it, my monitor goes black for a few seconds, then my desktop/Steam is back.

-There have been multiple other games in my library that did not perform well with this GPU; not sure if that's 'normal' or there is something else going on.

In any case, that was a metric fxck-ton of info.  If you read all of that, you have my gratitude.  This GPU has a 3-year warranty according to Newegg, from which I purchased it last April.

I know there are probably other steps I can still take to troubleshoot this...but I think if you need to made registry changes, and disable certain start-up programs, and this and that and the other thing, just to get games to play correctly...I mean as the customer I feel like that's not the way it should be lol.

Anyways, I'm grateful for any and all info / suggestions.  Thanks everyone,

Shaun

[Qoojo]

"Within 20 seconds of loading up a save file, my PC shuts off."

PC shutting down like that is a safety mechanism usually power, sometimes heat related. It sounds like you are narrowing in to heat issues with your PC. What sort of case are you using? I am guessing that you do not have good airflow into it.

Another thing to check is fan speed/fan %. If fans aren't spinning fast enough, it cannot cool as efficiently of course.

An observable side effect of thermal throttling is kind of what you are describing. Toggles between normal then slower FPS. When your gpu reaches a high enough temperature, it will thermal throttle. Which means it will clock down the GPU to run within thermal limits, thus lowering performance. Once it is cool enough, it will increase clock speeds again. So it's never ending cycle.

Generally these days hardware will downclock to combat high temps. So the PC shutdown could more related to power issue. But given your obvious heat issues, I wouldn't worry about it until you get the temps down.

Increase intake airflow to PC. Then adjust fan curves to be more aggressive if that does not help.

Also, note the difference between gpu temps and hotspot. If it is high enough delta, you might be able to RMA the GPU.

[sketterman]

Hi Qoojo,

Thank you for the reply, all good information.

1.  My case is a Phanteks Enthoo Pro ATX full tower.  Interestingly, I bought it after reading positive reviews about it regarding cable management and airflow.  I can't say for sure how good the overall cooling is, but I would argue that it has been fine for just over a year.  My thinking is if the case was too hot, this would have happened much sooner?  I did open the case up to see if it had gotten clogged up from dust, but it was surprisingly clean inside, top to bottom, front to back.

2.  Throttling is one thing I considered.  How fast, I wonder, can this theoretically happen?  In-game, when the issue occurs, it's ~ 1 second for a full cycle...so normal, slow, back to normal again in about 1 second, honestly it's probably a bit quicker than that.  I don't know enough about the technology in these GPU's to speak to this per se, but I would think the process would take at least a few seconds?  At least the 'cooling down' part of the cycle?

Another factor with this, is it doesn't appear to effect the entire game...just the player movement.  In Doom Eternal, for example, if I were to stand still in-game, and just observe the AI, everything moves smoothly and at normal speed.  It's only my movement that incurs the lag.

And then it's also that it doesn't happen in all games.

I'll see if I can find a way to 'test' my PSU.  If my PC shuts down while the PSU is stressed, but the GPU isn't, that would certainly be telling.

Or I can find a GPU-heavy test...but that simultaneously requires PSU load.

I can also contact ASRock tomorrow morning and explain what's going on to them.  

Shaun

[sketterman]

Post to add some more info:

Downloaded OCCT and ran a stress test.  Lasted about 3.5 minutes before my PC shut down.

Checked the Event Viewer...not much in there besides a general alarm that the PC shut down unexpectedly.  Google taught me that based on the details in the log file, that the shut down is occurring too quickly for the system to append any specific information.  So no dice there.

I went into my BIOS and configured every fan in my case to run at 100% indefinitely.  Then I downloaded and ran Fan Control, and ensured everything was running as such.

With the fans all at 100%, I launched Crysis Remastered.  GPU temps skyrocketed...core went to 103 C max, and hotspot was at 116 C max, but my PC didn't shut down.

With Adrenaline's metrics screen up on my 2ndary monitor, I watched the GPU temps as I alt-tabbed out of the game and back in.  As soon as I alt-tab, the temps plummet to the 60/70 C mark.  Clicking back into the game, they rocket back up to 100 C +.  The drop and raise of temps is nearly instantaneous.

But as soon as I go back into the game, I notice that my movement isn't quirky for a few seconds...then it starts to act up.  So I think the throttling issue is at least part of the problem.  But if my GPU is overheating with every fan in the case (including the GPU cooler fan) at 100%...and it didn't start doing this until a few weeks ago...that makes me think the GPU is failing.

On the other hand, I've never had a GPU overheat before...but I wouldn't think that would cause a complete PC shutdown.  That seems more like a PSU issue.  But I really have no clue.

I bought both from Newegg in the same week.  The PSU has a 10-year warranty, which is fantastic.  Newegg says the GPU has a 3-year warranty...though ASRock's website is ambiguous.  No doubt they'll point fingers at each other.  

In any case, any and all other suggestions are greatly appreciated.  Thanks again,

Shaun

[maverickwhites]

It looks like you’re facing multiple issues with your PC, mainly centered around high GPU temperatures and game performance problems. Here’s a structured approach to address these issues:

1. Address High GPU Temperatures

Your GPU temperatures are alarmingly high, which is likely the root of many performance problems:

Clean Your GPU and Case: Even though you mentioned that the case and filters are clean, dust buildup on the GPU’s heatsink or fans can cause overheating. Use compressed air to clean the GPU fans and heatsink.

Check GPU Fans: Ensure that the GPU fans are working properly. You can check this by observing the fans during gaming or using a tool like MSI Afterburner to manually adjust the fan speeds.

Improve Case Airflow: Ensure that your case has adequate airflow. Adding or repositioning case fans to improve airflow can help lower GPU temperatures.

2. Troubleshoot Game Performance Issues

The jittery performance and frame rate problems might be related to the high GPU temperatures or other settings:

Check for Driver Updates:Ensure that you have the latest drivers for your GPU from the [AMD website](https://www.amd.com/en/support).

Reinstall Graphics Drivers:Sometimes, drivers can become corrupted. Use the [AMD Cleanup Utility](https://www.amd.com/en/support/kb/faq/amd-clean-uninstall-utility) to completely remove old drivers before reinstalling the latest version.

Adjust Game Settings: Lowering settings or disabling specific features (like V-Sync) can sometimes resolve jittery performance. You can also try resetting the game settings to default. 
 3. Address System Stability Issues

The shutdowns during gameplay could be related to overheating or power issues:

Check PSU Cables and Connections: Ensure that all PSU cables are firmly connected and that there are no damaged cables.

Monitor System Temperatures: Use HWMonitor or similar tools to check CPU temperatures and other system components during gaming. Ensure that the CPU cooler is properly seated and that thermal paste is intact.
4. Seek Warranty Support

If the above steps do not resolve the issues, and especially if you find that the GPU is still under warranty:

Contact AMD or Newegg: Given that your GPU has a 3-year warranty from Newegg, you might want to contact their support for a potential RMA (Return Merchandise Authorization). They might offer repair, replacement, or refund options.

[sketterman]

Hi Maverick,

Thank you for the reply and all the great info!

I actually spoke to both ASRock and Thermaltake this morning, explaining to both of them what I was experiencing.  Thankfully, they both independently agreed that the GPU temps are way too high, and that the PC shutdown is likely due to a setting in my BIOS that auto shuts down the PC if any hardware reaches a certain temperature in order to prevent damage.

ASRock didn't ask too many questions, they just confirmed that the GPU was still within it's 3-year warranty period and sent me the documents to start the RMA process.  That has already been submitted to them.  So far, so good.

Thermaltake were really cool, too.  I didn't post this info here, but I ran the OCCT 'Power' stress test with my old GPU, and it ran fine for 30 minutes before I manually stopped it.  It gave me the average voltage for my entire PSU, and we went over those numbers:

+5V = 5.018V average
+3.3V = 3.267V average
+12V = 11.91V average
+1.05V PROC = 1.056V average
+1.05V PCH = 1.066V average
ATX +5VSB = 5.035V average

CPU (Intel 136000k) Core max temp = 99.5 C
GPU (NVIDIA GTX 1080) Core max temp = 84 C

He said my PSU is actually performing a bit better than advertised.  He also said that if I get a new GPU from ASRock, and the issue persists, to just contact them for a new PSU, easy day.

While nothing has been definitively proven, all signs are pointing to a faulty GPU.  It either has a defect in manufacturing that is causing it to overheat, or it might need to be repasted (which I'm happy to let them do).

Shaun

[maverickwhites]

Thanks for the update! It sounds like both ASRock and Thermaltake have been very helpful. I’m glad to hear the PSU is performing well, and I hope the RMA process resolves the issue. I’ll keep you posted on how things go with the new GPU.