Black screen crashes - think it's PSU?

[formulafox]

Hello, everyone, sorry to have to bother you with my troubles, but back in August 2012 I bought a prebuilt system from HP, the H8-1230 - specs here: https://support.hp.com/us-en/document/c03360575

I've not been running it in stock condition. I replaced the PSU with this: https://www.newegg.com/Product/Product.aspx?item=N82E16817182066
Replaced the GPU with a Sapphire Nitro Radeon R9 380
Replaced the stock CPU cooler with a Cooler Master Hyper 212 EVO
installed a 240GB SSD
And upgraded to Windows 10

Over the past couple of months I have occasionally the computer crash to a black screen. The event viewer records no errors, and the events recorded leading up to the crashes are never the same. After the most recent crash, Windows 10's Start Menu had reverted to it's default layout(no other settings changed, no programs disappeared).

I've been looking up a lot of information, including past threads on these very forums, and what I've read is leading me to think it's the PSU(and I hope it is - I can actually afford to replace that immediately) as the crashes have never happened while the GPU and CPU are under any real load. Most of them happen while I'm just doing something as simple as browsing the internet, and it's even happened while the computer is sitting and doing nothing while I'm at work.

The reason I'm posting in spite of what the information is leading me to think is because of one aspect of this that it's never clear is the case in previous threads; The system is still ON after these crashes, even ones that took place hours before I got home and found it, and I was under the impression that PSU failures cause the system to turn off.

So I was wondering if I'm thinking along the correct lines for this? Can this still be the PSU or should I be looking at something else? All my drivers and software, etc are up to date.

 

[norcalsc]

Have to believe it's not a gpu issue, but also psu hasn't been ruled out yet. Good luck

 

[formulafox]

(okay, what's going on? This makes THREE posts I've had to post TWICE before showing up!)

Okay, here's where we stand...

As a cause, I can rule out Malwarebytes. Two crashes with it turned off indicate that any software clash between MBAM and the AMD drivers are merely aggravating the problem, not causing it.

A friend of mine is planning to upgrade to a 1050 Ti and is taking my 380 to give a run in his(Very similar) system. He does not run Malwarebytes(he has BitDefender) so any software clash should be a non-issue - even moreso since he had an AMD processor already installed. I forget what the card is, but I recall him saying it's roughly equivalent to the GTX 750.

If I have no crashes, the 1050 Ti will be installed in three days - day one, no Malwarebytes just to be sure, day two Malwarebytes on but not set to start with Windows, day three Malwarebytes on and set to load with Windows(having that function on seemed guaranteed to cause a crash within minutes, which is what led me to think MBAM was the cause for a bit). If I have a crash and it comes up with the same dump information as before, the 1050 Ti will go in immediately and undergo the same verification process.

Speaking of the dump info, I never thought to check for dump files when a black screen went because I'd heard Windows doesn't know to create them. But I checked after the BLUE screen and found several dump files corresponding to my crashes. I have attached the most recent one in a Dropbox link should anyone want to look.

I looked at them in BlueScreenView and found that there are three files highlighted as a problem in every dump file: atikmpag.sys, dxgkrnl.sys, and dxgmms2.sys

ntoskrnl.exe and atikdmag.sys also appear on occasion. The former appears about half the time, the latter has appeared three times. They have never appeared together in any dump file. There are 18 saved dump files that actually have data - there are six more that are unreadable, all of the unreadable files dated before March 10th.

https://www.dropbox.com/s/if1p4iwl4hmkavo/WATCHDOG-20180312-0839.dmp?dl=0

EDIT: Almost forgot, but the earlier-mentioned errors in the event log were not present with the most recent two crashes.

 

[formulafox]

Would have posted this earlier, but had to leave for work almost immediately after settling this.

The 1050Ti is now in my system. It is now abundantly clear that the issue was indeed the GPU and the way it revealed itself as the culprit has made me decide I do not wish to ever put this thing back in unless my back is against the wall. Even then I would be doing it very hesitantly.

My friend immediately had trouble with the 380 not wanting to change resolution to the best suited for his monitor - black screen crash on each attempt. He brought it back and for due diligence I put it back in just to see. Initially worked as before, and for me it actually changed the resolution on request.

Then I turned on Malwarebytes. Remember when I suspected it to be the cause or a contributing factor? Well, there's definitely a software conflict aggravating the matter. Malwarebytes was turned on with the Start With Windows function(which I suspected might be the contributor rather than MBAM itself) turned OFF, and within three minutes, black screen.

And then I couldn't log in - whenever I opened the lock screen, black screen. If I waited ten seconds on the lock screen, black screen. Tried it five times. One of them got me a blue screen, with the same error as the prior blue screen occurrence.

Since MBAM wasn't starting with Windows, it couldn't have been triggering it. So out came the 380, in went the 1050 Ti.

And it got stuck in a loop of trying to autorepair and then going into the alternate startup menu. Wouldn't even load Safe Mode. System Restore has never worked no matter how many times I told it to turn on, so that was no go.

I've no idea how I fixed it - I had to go to the UEFI menu to check settings, but I changed nothing and that time it booted up fine. And booted fine after two restarts to ensure all was good.

So far, the 1050 Ti is running fine despite the initial troubles. MBAM is on and set to start with Windows, and it sat idle the entire time I was at work and has stayed on. Hopefully that will continue.

 

[Sedivy]

Hmmm. 380 should support most resolutions unless it's some really odd screen size. Could have been maybe a lack of drivers for the 380 on his rig.
I would however rma still, if you're within warranty, as such troubles with a gpu on two separate rigs aren't a good sign.

 

[formulafox]

When I first noticed this in January, I was six months past warranty. If I hadn't been, I'd have immediately RMAed the card for service and picked up a cheapo card so I could still keep in touch with friends while the 380 was checked.

As an update to my friend's efforts, he had thought of lack of driver updates as well so he made sure to update his software and tried the card again. He got it to work, but it hasn't gone more than about 18 hours without a black screen crash. He's picking up his own 1050 Ti tomorrow and he'll be done with the test of the 380.

As for me, I haven't had a single crash in the 5 days this card's been in my system - it did reboot unexpectedly once, but that was a Windows udpate that forgot it's supposed to ASK before restarting. The 1050 Ti is running everything I've asked it to without issue while running cooler than the 380 did. Anything this card isn't doing as well as the 380 is too small for my eyes to see.

 

[Sedivy]

Yeah that's pretty definitive. Hard to say exactly what the issue with the card is, but I wouldn't risk it anymore.
If this is now the third gpu you're swapping into your system, you have to wonder what's killing them off so quickly. If you're having heat issues (but it didn't sound like you were) or is it psu maybe having voltage variances outside tolerance levels for the cards. Did you end up checking the psu definitively?
https://www.wikihow.com/Check-a-Power-Supply
https://www.youtube.com/watch?v=ac7YMUcMjbw
Either way it's strange two of your gpus have died already.

 

[formulafox]

Technically I'm on the system's FOURTH GPU. The system came with a GT620, which died out within three months. I got a warranty replacement on that, which for some reason was the GT610 instead of the 620, but there was no discernible difference - PC was out of warranty when the 610 died. The 620 and 610 both died on the old PSU, as well - this is the first one I've lost on the new one.

The 620 and 610 both seemed to have junk coolers on them, so I think they just burnt themselves out.

 

[Sedivy]

The thing is 3 months is an awful short time for gpus to burn themselves out. They do so either out of high average heat condition, resulting in significantly shortened lifespan, or out of voltage issues which should really be affecting the whole system but likely show up on gpu first as it's the most power hungry component. In either case as gpu is the most expensive component in the system, I'd check your psu just in case, cause you don't want to keep wasting money on new ones. And maybe think about a new case/better cooling if psu turns out to be ok. I can almost guarantee it won't be though. PSUs in premade systems are given to cover just enough wattage as comes with the original components. If you're now inputting a gpu years newer, it'll be guzzling more watts, and your psu shouldn't be able to keep up. I'd be surprised if you don't immediately have issues.

 

[formulafox]

Everything I've seen says that the 1050 Ti pulls less less power than the 380(not surprising since I didn't have to hook it directly to the GPU and it's been running cooler according to Afterburner).

I'm not on the original PSU - I swapped it out as well when I got the 380. Went from the 350W it came with to a 550W.

 

[Sedivy]

Ah ok, different psu with higher wattage. I'd still check the voltages are within tolerances, just for your piece of mind. Something is making you burn through these very quickly and if it's not the psu, might want to think about investing in a better ventilated case.

 

[formulafox]

I'll be building a new system in June, and I will definitely be getting a new case. The tower it came with is so small that the Cooler Master Hyper 212 doesn't actually fit - I have to leave the side of the case off and make sure to clear the entire thing of dust on a weekly basis.

 

[Sedivy]

Ahh I see. Well if you're building a new one soon, then probably it'll last until then. Just make sure you check your psu before you try to reuse it in the new system.

 

[okcnaline]

Try another PSU and see if it fixes the problems.

 

[HEXiT]

okcnaline:

Nothing. There's no indication of something going wrong until it goes. If there's audio playing it right as the screen goes black it'll glitch out by looping about 1/4 second of the audio repeatedly for about five seconds before it goes quiet. After that, nothing can be done except to hold down the power button until the computer shuts off.

Sedivy:

I'll run that stuff and get back to you.

when i had this issue it was down to my gpu drivers and windows 7. the drivers kept crashing because i had hpet (high precision event timers) turned off... when i turned it on my system ran flawlessly...

as you upgraded to win 10 what o,s were you using previously?
if it was win 7 then go into bios and make sure hpet is disabled as win 10 doesnt use it and it can cause a black screen issue if its left on after upgrades.

also try rolling back your drivers to the previous version when your system was stable.

 

[formulafox]

okcnaline:

Nothing. There's no indication of something going wrong until it goes. If there's audio playing it right as the screen goes black it'll glitch out by looping about 1/4 second of the audio repeatedly for about five seconds before it goes quiet. After that, nothing can be done except to hold down the power button until the computer shuts off.

Sedivy:

I'll run that stuff and get back to you.

when i had this issue it was down to my gpu drivers and windows 7. the drivers kept crashing because i had hpet (high precision event timers) turned off... when i turned it on my system ran flawlessly...

as you upgraded to win 10 what o,s were you using previously?
if it was win 7 then go into bios and make sure hpet is disabled as win 10 doesnt use it and it can cause a black screen issue if its left on after upgrades.

also try rolling back your drivers to the previous version when your system was stable.

We're long past this point in fixing matters.

(though, for the record, the motherboard BIOS that this system uses, even after being updated to the latest BIOS, doesn't allow you to change ANYTHING)

 

[HEXiT]

is it a dell :/

 

[formulafox]

HP. And I'm as amazed as everyone else that this thing turned out to have any customizability - this was meant to be a short-term solution when my old built on a VERY tight budget PC fell too far behind to be viable. I'm very impressed that I've gotten nearly six years out of it - and it's still running well enough for me to preserve as an emergency backup when I build my new system.

 

[HEXiT]

that explains it then. hpet is likely locked on by default and because win 10 doesnt support hpet your having issues...
your only real option is to migrate back to win 7 as its the only o.s that supports hpet.

as you say your doing a new build you wont have to wait to long so your not gonna miss out on any games.... there aint many that are win 10 only, so yeah migrate back to 7 and your issues will likely stop.

 

[formulafox]

As I already noted, I didn't have any trouble after upgrading to Win10 until recently, and my friend hasn't been able to get the graphics card working either.

 

[HEXiT]

hmm ok