Help Me Diagnose This, Out of Ideas Now

[EXCellR8]

Decided that this deserved its own thread; I've mentioned this before somewhere (maybe in Vega builders) but it's still not fixed. So, here we go... sorry in advance for the length.

[system in question is first in my signature below, the Ryzen 1700 computer]

Started a few weeks ago while playing games, first time being Destiny 2 on the first release of Adrenalin drivers. Lost all video initially but then the sound bugged out maybe 10-20 seconds after; all I could do was reboot. Immediately rolled back drivers, making sure to do a clean install, but I think I may have used an older version of DDU. Long story short, the previous drivers (which I had been using just fine prior to updating) also crashed once the card started rendering 3D, or shortly thereafter.

So, naturally I uninstalled all AMD graphics drivers again and used the most recent DDU in safe mode to clean everything up. Didn't help, and the next go around the computer crashed just the same. Okay, so now I'm thinking the card is shot, so I remove it and try it in another machine, Also Win 10 Pro. Worked fine with latest drivers and didn't crash at all under heavy stress testing. A few suggested that it may be power related on the original computer but the PSU is actually 100W more than the minimum required for Vega. (I'm still trying to source another unit just in case this power supply is the culprit, but I don't think so).

Now, looking into other things I discovered my game drive had some unconfirmed sectors so I went and replaced the entire drive. System still crashes, updated some of my proprietary software (AIDA64, Logitech GS, even installed new sound drivers and confirmed DirectX integrity, system still crashes. Ran SFC on C: and confirmed that Windows installation was sound, updated and installed WhoCrashed, which didn't show any record of dumps, despite the Event Viewer showing that the display driver crashed but recovered (LIES!)

I'm pretty stumped at this point but I don't want to RMA yet because of how the card performed when installed in the other Windows 10 computer. This computer has been absolutely fine in the past, with a few little quirks that were easily ironed out. One of the only clues I have is that AMD drivers do NOT uninstall normally (i.e. progress gets to about 20% and just stops dead, never finishes), so that's odd but could be completely unrelated. Another clue is that all WattMan values default upon restarting but it's a hard restart so I suppose that's relatively normal.

I just flashed the BIOS on my board to the latest just for the heck of it and I've abandoned all OC values while trying to diagnose the issue. Nothing seems to have any affect on the problem though; pretty much all 3D apps and even some 2D stuff, such as Flash video and screen saver, crashes the computer. Looking at my keyboard LCD that runs the AIDA64 interface--once the screen goes dead the applet doesn't report stats for utilization or GPU temp, but the GPU dedicated memory still displays.

There aren't any new drivers for my board nor are there any Windows 10 updates available, so I'm not sure where to go at this point. If DDR4 wasn't so sinfully priced I'd try running a different kit but that's not happening. I still think it's a AMD driver bug but so far I've been unable to find the actual cause.

HALP?

 

[IRobot23]

1. PSU recommendations are nonsense. No you don't need need exactly 750W for VEGA system, or you might need even more. That depends on overclocking CPU/GPU. If you are going to going to get i7 7980X @4,5GHz even 1000W might be critical with super low power CPU.

2. Check PSU cables. My friend had same issue, cables made it crash.

3. Bug? Bug would show error not loss of power, but you never know.

 

[Iron Woode]

try a chkdsk run on the OS drive.

and/or

try a blank drive and a fresh win 10 install.

 

[EXCellR8]

Check PSU cables. My friend had same issue, cables made it crash.

3. Bug? Bug would show error not loss of power, but you never know.

I don't actually lose power, just video. The computer remains powered up after the screen goes dark but everything is still running.

try a chkdsk run on the OS drive.

and/or

try a blank drive and a fresh win 10 install.

I've already run chkdsk on the OS drive... but I'm thinking that maybe a regular OS update mucked something up. It's not likely but I suppose a bug could exist between Windows system and AMD's drivers. I can format another disk and give it a shot a bit later.

Thanks for the responses; I'll update as I go. Someone else is bound to encounter the same issue at some point.

 

[Iron Woode]

I have one more idea: run memtestx86 to make sure there are no ram issues.

 

[Schmide]

MSI Afterburner installed? I just had an issue where even if it wasn't running it messed with 17.12.2. Once uninstalled everything went back to normal.

 

[EXCellR8]

Nope, Afterburner is not installed. I used to use it religiously but stopped after Vega drivers kept breaking it. I think the last one I used was beta 19 but I had to remove it once AMD released a new set of software.

I suppose running memtest for some passes wouldn't hurt...

 

[Iron Woode]

Nope, Afterburner is not installed. I used to use it religiously but stopped after Vega drivers kept breaking it. I think the last one I used was beta 19 but I had to remove it once AMD released a new set of software.

I suppose running memtest for some passes wouldn't hurt...

you never know when a memory issue will pop up.

 

[EXCellR8]

The computer had been sitting idle and doing lightweight tasks for a few hours so I decided to run Cinebench. CPU test completed fine but the OpenGL test only lasted a few seconds before the system went down again. After rebooting I wanted to uninstall the graphics driver yet again but instead selected Modify > Clean Uninstall.

Seemed like it was actually going to work but it got to about 90% and then that crashed the computer also! Ughhh.

So, now I think I'm just going to purge all AMD drivers and try a different AMD card in this machine. That should eliminate the card as a causing factor for good; if the system crashes again, then I will have no choice but to rip it open and have a look at literally everything.

Thanks for the input; I will update with the results.

 

[Iron Woode]

cool. I hope you find the cause.

 

[EXCellR8]

thanks. I put in the RX 480 from my other Ryzen computer and it works fine with the latest Adrenalin drivers, no matter how hard I hammer on it. I'll just have to report to Sapphire tech support and see what they have to say about it. doesn't explain why the Vega card worked fine in the other machine but I haven't altered the card in any way so I should be able to get a replacement or refund. Shame, but what can you do?

thanks again for all the advice. I'll post back some time this week with Sapphire's official response.

 

[Reinvented]

I think your card is fine. Destiny 2 has a lot of problems with Vega cards. Stuff that would cause it to hard crash/lock up. Sometimes it's even down to the drivers, which I had problems on 17.12.1.

Wait till next patch. The Dawning stuff caused a bunch of problems for people.

 

[Iron Woode]

I think your card is fine. Destiny 2 has a lot of problems with Vega cards. Stuff that would cause it to hard crash/lock up. Sometimes it's even down to the drivers, which I had problems on 17.12.1.

Wait till next patch. The Dawning stuff caused a bunch of problems for people.

I think the card itself has some odd issue. It crashed trying to run the Cinebench OpenGL test.

 

[EXCellR8]

I think your card is fine. Destiny 2 has a lot of problems with Vega cards. Stuff that would cause it to hard crash/lock up. Sometimes it's even down to the drivers, which I had problems on 17.12.1.

Wait till next patch. The Dawning stuff caused a bunch of problems for people.

I hadn't had any issues with Destiny 2 since the initial Nessus crash months ago... the problem isn't game specific.

I think the card itself has some odd issue. It crashed trying to run the Cinebench OpenGL test.

The Cinebench OpenGL test barely puts stress on the card, so it's definitely not related to load or power delivery. I actually have the card in the other Ryzen 5 computer and it's been fine so far... I don't know. My gut still says that there's some kind of bug between the driver and something else I have configured in the other computer but I've not been able to isolate the cause, or even narrow it down much.

 

[VirtualLarry]

Is your Ryzen rig overclocked in any way? A "black screen crash" can happen due to OC. What's your SoC voltage?

 

[EXCellR8]

Yea both Ryzen machines are OC'd but I can also set everything default on the 1700 and it will still go down. Same deal with the most recent BIOS and default values, doesn't work no matter how many times you clear CMOS.

Also, I can run the same OC with the Polaris card and encounter zero issues... the problem is only present with the Vega. However, overclocking on the 1500X doesn't produce any problems with the Vega installed either, it's maddening.

 

[Iron Woode]

Yea both Ryzen machines are OC'd but I can also set everything default on the 1700 and it will still go down. Same deal with the most recent BIOS and default values, doesn't work no matter how many times you clear CMOS.

Also, I can run the same OC with the Polaris card and encounter zero issues... the problem is only present with the Vega. However, overclocking on the 1500X doesn't produce any problems with the Vega installed either, it's maddening.

that sounds like an OS issue.

did you try the blank drive in the vega system?

 

[EXCellR8]

Not yet, but I had suspected that an OS update may have broken it--which in that case I will likely have to wait for a new driver. The other Ryzen machine runs the same OS though so I will make note of what updates get installed going forward and what breaks it, if it happens on that system also. That would be a pretty solid indicator that there exists some kind of bug between AMD software and the OS itself--possibly due to optimization or power settings.

 

[Iron Woode]

Not yet, but I had suspected that an OS update may have broken it--which in that case I will likely have to wait for a new driver. The other Ryzen machine runs the same OS though so I will make note of what updates get installed going forward and what breaks it, if it happens on that system also. That would be a pretty solid indicator that there exists some kind of bug between AMD software and the OS itself--possibly due to optimization or power settings.

could be a driver for some other device, an app that isn't playing nice or even like you said an update to windows itself. That is why a blank drive would eliminate the existing install as the problem. If you can reproduce the issue, that could be submitted to AMD or Microsoft.

 

[EXCellR8]

Yes, I think I can officially rule out the card itself as the culprit because I cannot replicate the issue on two other Windows 10 computers. There are some apps that I've either updated or installed but so far nothing has helped the issue. I will certainly resort to a clean install if no future drivers and/or updates help my case.

 

[EXCellR8]

Trying a fresh install of the Windows 10 FCU and so far, so good, but I haven't done any robust 3D testing yet. Had an issue with some Visual Redistributiable package missing after OS installation but I think I've fixed that. Windows automatically installed some older Radeon Software but I'm running 17.12.2 now... hopefully no problems but I'll try some gaming tonight if we still have power. Supposed to be pretty snowy and windy for the day.

 

[Despoiler]

Trying a fresh install of the Windows 10 FCU and so far, so good, but I haven't done any robust 3D testing yet. Had an issue with some Visual Redistributiable package missing after OS installation but I think I've fixed that. Windows automatically installed some older Radeon Software but I'm running 17.12.2 now... hopefully no problems but I'll try some gaming tonight if we still have power. Supposed to be pretty snowy and windy for the day.

Automatic GPU drivers updates by Win10 is so annoying. I setup a group policy to disable updates for my GPU hardware IDs.

 

[EXCellR8]

Not a bad option... typically I wouldn't care but when I'm trying to troubleshoot having things automatically load up is super obnoxious.

No graphics related issues so far though, I must say.

 

[EXCellR8]

scratch that... crashed this morning whilst idle. it's gotta be the display driver itself.

still can't upgrade driver through Radeon Software without the whole system locking up.

 

[EXCellR8]

I think I may have figured this one out guys... but I won't know until I have a new power supply and cables.

Upon close inspection it looks like, and I don't know for sure, but the card may have been drawing so much power from a single PCIe power connector on the PSU that it actually fused the end of the cable into the port! I couldn't get the cable out, not even with pliers! The end of the cable actually snapped right off with two teeth still stuck in the power supply board so I think that might be what's up--or that once the card starting to draw heavy power the supply was cutting it off if there was a spike or short or something. Gah, that doesn't seem possible but needless to say I think I'll leave the power limit alone and use two separate cables with the new power supply!