Black screen crash GPU 100% fan - Weird issue

[shankly1985]

Hello I have a very strange issue happening with my PC atm, first let me give you the specs

i7 8700k - Tested OC and Stock - Running Stock atm
ROG Maximus X Hero
16GB Corsair Dominator 3200mhz tested at the XMP speed at Stock - Running stock atm
1000w Superflower leadex platinum
RX VEGA 64 - Stock

This only happens when running games, its never happened when I just idea or using Chrome etc At random my screen with just turn off and the GPU fan speed will boost to 100% and audio will repeat itself i need to hold power button to switch off. When I back onto the desktop and I look inside Event Viewer I see that the Display driver had stopped Responding and Recovered. It never Recovers
So at this stage I start running all tests, RAM, SSD, HDD, changing PCIe cables maybe its the PSU? CPU? etc

This crash is so random that I have had it happen 30 mins into a game and 12 hours I had a game running for Alt-tabing in and out for 12 hours till it crashed.

I believe I found the issue and so far I not crashed for almost a day now. If I leave RX VEGA Wattman on balanced preset it will black screen
What I did was setup a custom wattman set the power limit to 50% and adjust the core clocks abit to 1300/1400 about what I get with balanced so it really hasn't changed....

Does anyone know why a stock preset from AMD Balanced mode might crash a system?

 

[Sasqui]

I believe I found the issue and so far I not crashed for almost a day now. If I leave RX VEGA Wattman on balanced preset it will black screen
What I did was setup a custom wattman set the power limit to 50% and adjust the core clocks abit to 1300/1400 about what I get with balanced so it really hasn't changed....

Does anyone know why a stock preset from AMD Balanced mode might crash a system?

Update your system specs! ...that's quite an upgrade

I don't know specifically, but can tell you after owning 2 Vega 64's and playing for hours with both, they behaved fine. I did notice the second one (they were both Powercolor), it ran almost 15c hotter than the first. My guess is that there's some instability with that particular card in the stock mode (overstating the obvious). Have you measured peak core/mem speeds and temps with GPUz???

 

[shankly1985]

Update your system specs! ...that's quite an upgrade

I don't know specifically, but can tell you after owning 2 Vega 64's and playing for hours with both, they behaved fine. I did notice the second one (they were both Powercolor), it ran almost 15c hotter than the first. My guess is that there's some instability with that particular card in the stock mode (overstating the obvious). Have you measured peak core/mem speeds and temps with GPUz???

Yeah nothing looks out of place, core speed starts of around 1500mhz memory 945 the core speed drops down and remains around 1300mhz has the GPU heats up. the max temp is no more than 80c on the core and memory.

 

[Melvis]

When ive seen this before from AMD cards, it ends up to be faulty, try different drivers if that doesnt work id say RMA the card

 

[Komshija]

If your GPU from Sapphire? I don't know whether such issue affects all AMD cards, but similar problem affects my Sapphire Nitro R9 390 - whenever I modified frequency or voltage values. My screen doesn't crash or anything like that, but I experience massive throttling and when I exit the game and run TRIXX, fans automatically spin to 100%. The GPU was overheating because TRIXX interfered with cooling and the fans didn't spin.
I found a solution! I uninstalled TRIXX, removed its remains manually from the PC and registry, restarted the PC and installed TRIXX again - just for the sake of custom fan curve. I advise you to avoid overclocking, undervolting, overvolting, raising power limits etc. Or if you do, set custom fan curve or fixed fan speed - eg. 60%.

I advise you to reset the GPU to stock values, uninstall Radeon software, remove its remains (because software never uninstalls completely), restart the PC and make a clean install of the same software. Do not modify anything and see if the problem persists. If it doesn't, the card isn't faulty.

 

[shankly1985]

If your GPU from Sapphire? I don't know whether such issue affects all AMD cards, but similar problem affects my Sapphire Nitro R9 390 - whenever I modified frequency or voltage values. My screen doesn't crash or anything like that, but I experience massive throttling and when I exit the game and run TRIXX, fans automatically spin to 100%. The GPU was overheating because TRIXX interfered with cooling and the fans didn't spin.
I found a solution! I uninstalled TRIXX, removed its remains manually from the PC and registry, restarted the PC and installed TRIXX again - just for the sake of custom fan curve. I advise you to avoid overclocking, undervolting, overvolting, raising power limits etc. Or if you do, set custom fan curve or fixed fan speed - eg. 60%.

I advise you to reset the GPU to stock values, uninstall Radeon software, remove its remains (because software never uninstalls completely), restart the PC and make a clean install of the same software. Do not modify anything and see if the problem persists. If it doesn't, the card isn't faulty.

The GPU is a Ref Sapphire. I dont install any OC software like Trixx or Afterburner. This is happening on the out of the box settings I have even couple days ago reinstalled Windows 10 full fresh install and just had the basic drivers and a game installed for testing the issue remained.

My only fix so far is changing the core states myself to a lower number than what is out of the box. So far I am yet to crash doing this.

Am just confused by how this is fixing the problem. I still getting the same speeds now vs the balanced preset.

When ive seen this before from AMD cards, it ends up to be faulty, try different drivers if that doesnt work id say RMA the card

I think you right, or my PSU is the issue.

 

[hyp36rmax]

I've had this issue with a few VEGA 64's on one particular system with a Ryzen 1700X, ASUS Crosshair Hero VI, and Seasonic 1200 Watt XP3 Platinum and 1000 Watt PRIME Platinum PSU's. Oddly once i Transferred my crossfire VEGA 64's to a system with an Intel 8700K, ASUS Maximus Hero X, and EVGA 1600 Watt T2 Titanium PSU, all those issues went away. Still having an issue with this on a VEGA 64 LE on my AMD setup.

 

[eidairaman1]

If you lowered gpu core states and it is fine, replace the card.

How old is the psu?