whea_uncorrectable_error in gaming or GPU test

[Vya Domus]

Using HWmonitor for voltage monitoring is not a route you want to take.

Why ? CPU voltages were always spot on for me on every computer I've tried it on.

 

[Renan]

Can't really determine anything yet , they do seem bit low tho, keep HWmonitor on and run the benchmark tool in CPU-Z and post another pic of this.

Bench:

Stress:

 

[Vya Domus]

CPU voltage is too low in my opinion, I think that's causing your problem , you can go into bios and manually set your CPU voltage to 1.35V and be sure Turbo Core is off even though it seems it already is. You might need to turn OC mode on or whatever these settings are named in your bios. But be careful what exactly you change , make sure all failsafe options are enabled.

 

[Renan]

CPU voltage is too low in my opinion, I think that's causing your problem , you can go into bios and manually set your CPU voltage to 1.35V and be sure Turbo Core is off even though it seems it already is. You might need to turn OC mode on or whatever these settings are named in your bios. But be careful what exactly you change , make sure all failsafe options are enabled.

So I go in the BIOS, put 1.35v manually and run Valley Benchmark?

 

[Vya Domus]

That or whatever else caused those crashes before.

 

[Renan]

Put 1.475v in the BIOS.

HWMonitor, CPUZ and AIDA64 shows 1.320v

Run Valley Benchmark, crash after 3 minutes.

 

[cdawall]

Why ? CPU voltages were always spot on for me on every computer I've tried it on.

You are betting that a software sensor works correctly to read voltage of an external component it has no control of.

 

[Vya Domus]

You are betting that a software sensor works correctly to read voltage of an external component it has no control of.

I am betting on that fact that it gives a reading precise enough to at least get a general idea , in lack of anything better.

 

[Vya Domus]

Put 1.475v in the BIOS.

HWMonitor, CPUZ and AIDA64 shows 1.320v

Run Valley Benchmark, crash after 3 minutes.

1.475 is too high , set it back to auto or 1.35. So whatever that error is about it doesn't have anything to do with the CPU. The fact that even after going back to an older driver didn't solve the problem it means the GPU is dying or it's the PSU. Do test the GPU in a different system if you can , otherwise I don't how I can further sort this out.

 

[cdawall]

I am betting on that fact that it gives a reading precise enough to at least get a general idea , in lack of anything better.

It does or does not depending on board judging from the supremely inexpensive board he is using I would say it does not.

 

[Vya Domus]

It does or does not depending on board judging from the supremely inexpensive board he is using I would say it does not.

We work with what we have , if it was as easy as just buying something more expensive/better we would've been done long ago.

 

[OneMoar]

@Vya Domus hush you don't have a clue what you are talking about
!op
your temps are what I would consider warmer then what the fx chips generally tolerate remember the sensor you are reading is the socket sensor NOT the core the core is likely well over 70c
FX chips REALLY don't like to be over 60c at the core level

stock voltage for a 6300 is about 1.225v on later samples

I would consider that board suspect tho its a old 780lm board they where not very good when they where new and they are pretty crappy theses days I would't use one in a system ever... especially the gigabyte ones

get some more cooling on the board and run memtest overnight
and upload your minidump so we can look at it

 

[Vya Domus]

@Vya Domus hush you don't have a clue what you are talking about
!op
your temps are what I would consider warmer then what the fx chips generally tolerate remember the sensor you are reading is the socket sensor NOT the core the core is likely well over 70c
FX chips REALLY don't like to be over 60c at the core level

Sorry but YOU don't know what you are talking about , for FX cpu's core/package temps are always below socket temps .

 

[cdawall]

Sorry but YOU don't know what you are talking about , for FX cpu's core/package temps are always below socket temps .

Depends on board...

Why are you arguing with experienced people trying to help? Seriously if you can't take advice figure it out on your own. I'm out.

@eidairaman1 @erocker @Fourstaff @Mussels @Paulieg

 

[OneMoar]

when you have a system then suddenly goes from being stable to not IT NOT HAVE ANYTHING THE FAWK todo with bios settings

are you seriously gonna tell me that there where gremlins running around in the middle of the night changing bios settings

NO thats now how technicians troubleshoot,thats how people that's how people that don't have a clue about how this shit accually works trouble-shoot (well I don't know whats really going on so I am just gonna tell people to push all the buttons until something happens) NOPE
thats how stuff gets screwed up which hampers the trouble shooting process for people that do know


the northbridge or the vrms on the board are could be on there way out I have burned up/seen burned up more 780 Chipset boards then I care to count,psu faults generally present as a hard-reset its very rare these days for a psu to cause general instability unless you buy a super-crappy unit,Cpu instability is 80% of the time a STOP 0x0xxxxxxxx124,memory errors also usually present as a paging related error


the 280X gpus are OLD just last week I threw 4 of them in the trash at work because they had died of old age

they
where not exactly reliable infact one of the early symptoms of a failing 7970/280x are a WHEA error shortly after that it progresses to a corrupt or black screen then it quits

if your are getting a screen with vertical colored lines then the card is dead that is the defacto failure-mode for a Tahiti XT card

need to see the full minidump to get a better idea of what was in memory at time

the op has multiple issues going on here the first and foremost is that 280x is probably dying
secondly the motherboard is a utter pile and probably contributing to the funky error message instead of the classic screen-crash clearly indicating a dead gpu

 

[Renan]

So I tried something different.

I reset the BIOS again.

Lowered de GPU clock from 940mhz to 900mhz, and GPU memory clock from 1250 to 1200mhz on RADEON SETTINGS.

I got a big "fan", I don't know how to call it in English, here in Brazil we call it "ventilador".

I put it next to the case that was already open, like this: (not my picture):
I also removed the top cover of my Cougar MX300.

I opened the NBA 2k17, which is the fastest way to crash. Put everything on Medium/OFF in the Settings, I left in 1080p.

I was able to play 2 whole quarters without any crash before close the game by myself.

60 FPS
The temperature of the GPU never exceeded 55 ° C.
CPU does not exceed 35 ° C
GPU and CPU usage did not exceed 45%.

Does that mean anything?

 

[OneMoar]

gpu is on its way out bud

we are done here
./thread

 

[Vya Domus]

Depends on board...

Why are you arguing with experienced people trying to help? Seriously if you can't take advice figure it out on your own. I'm out.

@eidairaman1 @erocker @Fourstaff @Mussels @Paulieg

Because I have experience of my own , I've build several AMD systems and OC'ed them I have good understanding of them and I know those temps aren't dangerous. I ain't arguing with no one , just stating some known facts.

 

[Vya Domus]

So I tried something different.

I reset the BIOS again.

Lowered de GPU clock from 940mhz to 900mhz, and GPU memory clock from 1250 to 1200mhz on RADEON SETTINGS.

60 FPS
The temperature of the GPU never exceeded 55 ° C.
CPU does not exceed 35 ° C
GPU and CPU usage did not exceed 45%.

Does that mean anything?

The GPU is on it's way out most likely .

 

[Renan]

So, I put NBA 2k17 on Max Settings like I always played.
Play a whole quarter, no crash. 60 FPS
GPU temp stabilized in 65 ºC (with that big fan in my Case).

Do you think I should buy a good thermal paste and replace the CPU and GPU paste or give up of my r9 280?

 

[cdawall]

Because I have experience of my own , I've build several AMD systems and OC'ed them I have good understanding of them and I know those temps aren't dangerous. I ain't arguing with no one , just stating some known facts.

You know nothing in the grand scheme of things remember there are members on here who have thrown away more AMD systems than you have even thought about in your wettest of dreams.

 

[Vya Domus]

You know nothing in the grand scheme of things remember there are members on here who have thrown away more AMD systems than you have even thought about in your wettest of dreams.

The cringe is real, what can I say.

 

[eidairaman1]

I can show dangerous temps in a bit.

 

[Renan]

Sorry to interrupt your discussion about voltage.

Continuing...

I took out that big fan I had placed throwing air into the system.

I played NBA 2k17 for 20 minutes without crash. (Remember, the game crashed after about 20 seconds after the first jumpball).

GPU temp stabilized in 65 ºC while I playing (same as before the big fan).

So... now the only thing I did was lower the GPU clock from 940mhz to 900mhz.

That made me think ...

The standard clock of a r9 280 is 827mhz.
My Sapphire r9 280 came with 940mhz from the factory.

Do you guys think taht after so much time with this boost clock as default it has become defective and now only works with a lower clock?
Would it be possible?

 

[jboydgolfer]

Do you think the problem is not the PSU?
Like, if I run the test on the onboard video and work fine, can't it be the PSU that can't provide enough power for the whole system when it's in full load?

Yes, I have onboard video.

With which program do I test the integrity of the GPU? Because I run Prime95 for one hour (with GPU plugged) and did not crash.

What I was getting at is this. By removing the discrete video card and verifying the integrity of the processor, you would remove the processor being a problem out of the equation. Prime 95 is a perfectly fine CPU test

Then once verified with the CPU, you would reinstall the discrete video card and stress it, either with benchmarks of which there are countless numbers available on the Internet for free and Here tech power up as well. Or you could just simply game with it. If it crashed you would know that it was either your video card or your power supply of course those arent the only possibilities but you would have removed the processor from the equation as being a possible culprit.

A PC has so many components that can fail that just saying you're having a problem, and excepting suggestion after suggestion doesn't really get you anywhere to a resolution. My recommendation is using the process of elimination and common sense and arriving at an answer that way

an example from my own life is as follows....
i recently built a PC for my nephew, I installed all the parts, and it wouldnt fire up. what i then did was swap the PSU, I pulled the one installed, and put in another one that i got from a friend, the PC fired Right up, Just fine after replacing the PSU. Before swapping PSU tho, i pulled RAM, Reseated the CPU, and GPU, and THEN arrived at trying the PSU. That is the best method in my experience whe ntroubleshooting an issue that You ar not certain of what is causing it.