PSU or gpu dying?

[fixME]

Woke up the other day to find my computer with no display.  I took off the side panel and noticed the gpu fan was not spinning at all.  Prior to this, the computer was working fine.  I was gaming the night before, then went to sleep for 8 hours and the display was black when I woke up.  There wasn't much time for troubleshooting since I had to leave for work.

 

 

I just came home to check on the computer which I left running.  I looked for signs of damage, tried unplugging/replugging the DVI cable,  switching the monitor inputs etc, but no go.  So I removed the graphics card and restarted the computer.  It booted fine using onboard HD 4000.

 

My psu is an XFX 550w Bronze Seasonic OEM which I got two years ago.  It still has 3 years on a five year warranty.

My gpu is a Powercolor 7870 PCS+ (7870 Tahiti LE) which I also bought two years ago.  The warranty expired less than 20 days ago.

 

Nothing is overclocked in my system right now.  I am running a 3570k, 3 SSDs, DVD drive, one mechanical drive, and the 7870.  All that can be powered by the XFX 550w.

 

Unfortunately I don't have another psu right now to test the graphics card.

Suggestions?

 

 

 

 

[Skeletonfish]

im thinking its the gpu just because if it was the PSU other things relying on the 12v rail would be going wonky 

[CalculuS]

Test prime95 while using onboard to make 100% sure that PSU is still capable.

[fixME]

Test prime95 while using onboard to make 100% sure that PSU is still capable.

I just ran the 'In-place large FFTs (maximum power consumption) for 35 minutes.  All tests passed, no errors.

 

So you believe it's the gpu?

[CalculuS]

I just ran the 'In-place large FFTs (maximum power consumption) for 35 minutes.  All tests passed, no errors.

 

So you believe it's the gpu?

Yup.

 

If you want to be 100% sure, test it in another PC.

[Phresh]

Does the card use an adapter from the Power Supply if so test it with a multi-meter i have had issues in the past with the one that came with my Asus GTX 760.

[fixME]

Yup.

 

If you want to be 100% sure, test it in another PC.

I will when I get the chance, but I believe you are right.

 

I ran the Small FFTs for 20 minutes and everything passed.

 

I like how the card gives up just as the warranty expires.  I doubt Powercolor will resolve anything for me, but I'll contact them anyway.

 

 

Does the card use an adapter from the Power Supply if so test it with a multi-meter i have had issues in the past with the one that came with my Asus GTX 760.

No adapter.  The psu has 3 x 6+2-Pin.

[faziten]

A 3570K wont stress enough a seasonic 550W to determine if its a psu issue.

My suggestion is:

1)Connect the gpu (AMD), plug the PCI lines, but boot with your integrated H4000. (You may prior to this configure your BIOS to let you boot with integrated graphics even in the presence of a PCI device).

2)Start pc and see if AMD GPU is recognised (Device manager > Display adapters). If it is, update driver (if already updated to last, rollback to a previous one)

3) Restart with monitor pluged in integrated graphics.

4) while on windows plug the monitor into AMD GPU output, see if works.

 

If all of this fails Go back to step 1, changing the PCI cables (if your PSU doesn't have more, try molex to PCI or viceversa you should have some on your GPU box, they all come with a pair of molex to PCI.).

[fixME]

A 3570K wont stress enough a seasonic 550W to determine if its a psu issue.

My suggestion is:

1)Connect the gpu (AMD), plug the PCI lines, but boot with your integrated H4000. (You may prior to this configure your BIOS to let you boot with integrated graphics even in the presence of a PCI device).

2)Start pc and see if AMD GPU is recognised (Device manager > Display adapters). If it is, update driver (if already updated to last, rollback to a previous one)

3) Restart with monitor pluged in integrated graphics.

4) while on windows plug the monitor into AMD GPU output, see if works.

 

If all of this fails Go back to step 1, changing the PCI cables (if your PSU doesn't have more, try molex to PCI or viceversa you should have some on your GPU box, they all come with a pair of molex to PCI.).

I set to integrated in bios, shutdown the computer, and plugged in the gpu.  I then pressed the power button and the gpu fan spun for a second, then turned off.  The display remained black and then the computer restarted itself after two seconds and the gpu fan started spinning.  I am in Windows now and the graphics card is detected in device manager.

 

About the driver though.  It says I am on the latest version.  However, a few weeks ago Catalyst said there was a new driver update whenever I restarted my computer.  I just ignored it and said 'remind me later.'  So device manager is wrong, there is a new driver.

 

I cannot update through Catalyst control, probably because I using integrated.

 

On AMD's website it says the latest is 'Omega (14.12)', released 12/8/2014.

 

Under device manager for me it says 14.100.0.0, Catalyst control says Direct3D 9.14.10.01029, OpenGL 6.14.10.12874.

 

What is your suggestion?  Roll back the driver in device manager or download the new driver from the AMD website?

[faziten]

I set to integrated in bios, shutdown the computer, and plugged in the gpu.  I then pressed the power button and the gpu fan spun for a second, then turned off.  The display remained black and then the computer restarted itself after two seconds and the gpu fan started spinning.  I am in Windows now and the graphics card is detected in device manager.

 

About the driver though.  It says I am on the latest version.  However, a few weeks ago Catalyst said there was a new driver update whenever I restarted my computer.  I just ignored it and said 'remind me later.'  So device manager is wrong, there is a new driver.

 

I cannot update through Catalyst control, probably because I using integrated.

 

On AMD's website it says the latest is 'Omega (14.12)', released 12/8/2014.

 

Under device manager for me it says 14.100.0.0, Catalyst control says Direct3D 9.14.10.01029, OpenGL 6.14.10.12874.

 

What is your suggestion?  Roll back the driver in device manager or download the new driver from the AMD website?

Mmm... odd what you say about it starting and stopping.

 

I'd attempt a driver rollback, if cant or just doesn't work, try to update to last driver (download manual from AMD site and execute it).

If fails, follow the solving path any technician would (i do it myself for my clients):

 

1) Discard Hardware: (jump to next step only if present is OK).

Remove GPU from your pc.

---a) Grab an electric tester if you can and test V12 lines in your plugs. Should be 12v +/- 0.5v. IF not available, enter BIOS and search for the tab "satus" or "pc health" and see if voltages reported show anomalous numbers 12V should be close to 12. (if you cant find it tell us de MoBo brand and name).

---b) Unplug the GPU PCI lines and see if the connector presents any object that may be interfering with metallic contacs (also see if the plug is broken or bent anomaly).

---c) Visually check if GPU PCB is bent (land it in some even surface or use something straight -ruler-) 6/7000 series are known to bend with time if longer than 10 inches.

---d) Check if GPU power connectors have play (do not stress them, if they have play they should move easily otherwise not).  Also check if power connector presents objecs inside it like paper,dust, etc.

---e) connect the GPU back into your MoBo. See if PCI lines are making full contact, to see this, the best way is watch the PCB of your GPU going in the PCI express in your MoBo. You shouldn't be able to see golden conectors. (They must be all covered by the Socket to assure the card is fully in). 

***Do you have a speaker installed? the tiny thing that should beep when you turn on your pc. If have plug it in, if already pluged in, How many beeps does it when you turn on?***

Boot pc and see if works. If not, go to step 2.

 

2) Discard Firmware:

---a) Boot your PC into bios with whichever video adapter works: Go to "exit" menu and roll back your bios to stock values. Reboot, and try connecting your monitor to the GPU. See if works. If you know how, use your MoBo jumpers to return stock values. 

 

3) Discard Software: (at this point would be very uncommon).

---a) Download any linux distro, burn it to DVD or Pendrive and boot into linux to see if it works.

---b) If your gpu does not display video feed, start a console and type: "dmesg" without quotes. Press enter and search for the words "gpu" "amd" "display" "pci" or just copy paste here.

 

If everything fails, ask someone to lend you a PSU. 

If still not working, Take your GPU and test it into another Motherboard/Pc (make sure its PSU can handle power requirements).

 

If the card is not working despite all of this... I'm out of suggestions and will asume the GPU has died.

 

If really want to find the source of the problem a P.O.S.T. PCI card should give a number indicating the problem. But i assume you don't have one and wont spend the money buying one. Also, note some motherboards have one included and will show a pair of alphanumeric characters that can be decoded into reports.

 

That's what i would do with my clients and my own hardware. 

Best of lucks solving!

[fixME]

Mmm... odd what you say about it starting and stopping.

 

I'd attempt a driver rollback, if cant or just doesn't work, try to update to last driver (download manual from AMD site and execute it).

If fails, follow the solving path any technician would (i do it myself for my clients):

 

1) Discard Hardware: (jump to next step only if present is OK).

Remove GPU from your pc.

---a) Grab an electric tester if you can and test V12 lines in your plugs. Should be 12v +/- 0.5v. IF not available, enter BIOS and search for the tab "satus" or "pc health" and see if voltages reported show anomalous numbers 12V should be close to 12. (if you cant find it tell us de MoBo brand and name).

--- B) Unplug the GPU PCI lines and see if the connector presents any object that may be interfering with metallic contacs (also see if the plug is broken or bent anomaly).

---c) Visually check if GPU PCB is bent (land it in some even surface or use something straight -ruler-) 6/7000 series are known to bend with time if longer than 10 inches.

---d) Check if GPU power connectors have play (do not stress them, if they have play they should move easily otherwise not).  Also check if power connector presents objecs inside it like paper,dust, etc.

---e) connect the GPU back into your MoBo. See if PCI lines are making full contact, to see this, the best way is watch the PCB of your GPU going in the PCI express in your MoBo. You shouldn't be able to see golden conectors. (They must be all covered by the Socket to assure the card is fully in). 

***Do you have a speaker installed? the tiny thing that should beep when you turn on your pc. If have plug it in, if already pluged in, How many beeps does it when you turn on?***

Boot pc and see if works. If not, go to step 2.

 

2) Discard Firmware:

---a) Boot your PC into bios with whichever video adapter works: Go to "exit" menu and roll back your bios to stock values. Reboot, and try connecting your monitor to the GPU. See if works. If you know how, use your MoBo jumpers to return stock values. 

 

3) Discard Software: (at this point would be very uncommon).

---a) Download any linux distro, burn it to DVD or Pendrive and boot into linux to see if it works.

--- B) If your gpu does not display video feed, start a console and type: "dmesg" without quotes. Press enter and search for the words "gpu" "amd" "display" "pci" or just copy paste here.

 

If everything fails, ask someone to lend you a PSU. 

If still not working, Take your GPU and test it into another Motherboard/Pc (make sure its PSU can handle power requirements).

 

If the card is not working despite all of this... I'm out of suggestions and will asume the GPU has died.

 

If really want to find the source of the problem a P.O.S.T. PCI card should give a number indicating the problem. But i assume you don't have one and wont spend the money buying one. Also, note some motherboards have one included and will show a pair of alphanumeric characters that can be decoded into reports.

 

That's what i would do with my clients and my own hardware. 

Best of lucks solving!

Ok so I updated to the latest AMD omega driver.  I did as you said, restarted, and then connected DVI to graphics card upon entering Windows.

 

So far it looks like its working.

 

Idle temps:

 

GPU temp is 32 degrees Celsius

GPU usage is 0

Fan speed is 25%

 

Playing CS:GO for 10 minutes

 

GPU temp around 66

GPU usage is 70-90

Fan speed is 35%

 

Should I go through any of the other steps or continue to use it?

Next weekend I'll have the opportunity to test it another computer.

[fixME]

Everything seemed to be working until I woke up again this morning.  The computer had restarted itself and Windows had 'recovered from an unexpected shutdown' -   Problem Event Name:    BlueScreen.

 

So perhaps the gpu is slowly dying.

What do you think?

[faziten]

Sorry the long time no answer;

Imho, get rid of it. Out of guarantee, randomly failing, BSODing your entire system. Simple: sell it.

 

Now, to be sure follow the other steps just to know is not your PSU or MoBo failing on you. 

Everything seemed to be working until I woke up again this morning.  The computer had restarted itself and Windows had 'recovered from an unexpected shutdown' -   Problem Event Name:    BlueScreen.

 

So perhaps the gpu is slowly dying.

What do you think?

[fixME]

Sorry the long time no answer;

Imho, get rid of it. Out of guarantee, randomly failing, BSODing your entire system. Simple: sell it.

 

Now, to be sure follow the other steps just to know is not your PSU or MoBo failing on you. 

I was just watching Netflix when all of a sudden the video froze and the sound kept looping the last 15 seconds.

I managed to escape out back to desktop and the sound stopped, but I could not close the program.  Alt+ctrl+delete did not work and I could not close anything via taskbar.

 

I was able to alt-tab through programs but it wouldn't let me close any.  Eventually a few of them ended up freezing.

 

Clicking the taskbar got me the popup 'The application is not responding.  The program may respond if you wait.  Do you want to end the process?'  End the process did nothing, cancel did nothing, and 'X'ing' the window did nothing.  Every time I clicked the taskbar I got that popup.  I waited 1 hour 15 minutes and the situation did not resolve it self.

 

Pressed the power button, waited a few seconds and rebooted.  GPU fan made noise, but black screen.  Rebooted again, same thing.  Plugged into the onboard graphics and the computer boot.

 

Would you say this is another confirmation that the gpu is on it's way out?

[faziten]

I was just watching Netflix when all of a sudden the video froze and the sound kept looping the last 15 seconds.

I managed to escape out back to desktop and the sound stopped, but I could not close the program.  Alt+ctrl+delete did not work and I could not close anything via taskbar.

 

I was able to alt-tab through programs but it wouldn't let me close any.  Eventually a few of them ended up freezing.

 

Clicking the taskbar got me the popup 'The application is not responding.  The program may respond if you wait.  Do you want to end the process?'  End the process did nothing, cancel did nothing, and 'X'ing' the window did nothing.  Every time I clicked the taskbar I got that popup.  I waited 1 hour 15 minutes and the situation did not resolve it self.

 

Pressed the power button, waited a few seconds and rebooted.  GPU fan made noise, but black screen.  Rebooted again, same thing.  Plugged into the onboard graphics and the computer boot.

 

Would you say this is another confirmation that the gpu is on it's way out?

that kind of issues make mi wish think it's a driver thing. -But you've been through there and nothing changed so keep it simple: Sell it, buy new one.