Overheating CPU and GPU

[Tomasoff]

Hello everyone,

I have my computer for about 2-3 years, it never had problems, but since 2014 12 25 it started to shut down during gameplays (DotA 2 and CS:GO). At first, I thought it was just an error, but then it happened a few more times. I downloaded speedfan and saw that during gameplay, my GPU reaches 83 C and CPU ~70 C. I opened the pc case and put a big FAN close to the components and set it to max. It helped a little bit, at least PC didn't turn off, but today it is not helping anymore, so I went and bought some Thermal Paste (CoolerMaster E1 IC Essential). I watched a video how to change it, I went with the "dot" on CPU and the plastic bag style with my GPU. CPU seems to have 50-60 C during gameplays, but GPU still reaches 83C WITH EXTERNAL FAN ON! I don't know what to do anymore. Please help!

Intel Core i3 2120
nVidia GeForce GTX 560
8GB of RAM
1TB

Tomas

 

[Carobthomp]

It's possible that for whatever reason the air is not cycling through the tower as it once did. It may be worth while searching for additional exterior cooling options or even moving the GPU to a lower slot on the MB if there is room on the mother board.

Alternatively you could take a look at the GPU itself, it may be time for some gentle dusting in case the fans on it have built up an access amount of dust.

 

[Tomasoff]

I cleaned the whole computer before applying the Thermal paste.

 

[madmatt30]

83c is not particularly hot for a gtx 560 under heavy load - 95c+ is a dangerous temp ,anything under 90c is acceptable.

 

[Tomasoff]

So why the pc is turning off then? :/

 

[Carobthomp]

Have you added any additional hardware since this started happening? Is it possible it's now more of a PSU issue?

 

[Tomasoff]

I only changed external hardware: new mouse, monitor, keyboard, headphones :/

 

[Tomasoff]

help?

 

[madmatt30]

whats your psu model mate,Im looking at this as the main culprit now personally

 

[Tomasoff]

whats your psu model mate,Im looking at this as the main culprit now personally

http://gyazo.com/29ba302bd8cc896f282665ca1d6da33e

 

[Tomasoff]

Speccy also shows this: http://gyazo.com/10fd16134ccd1cf87fffa1b56b262513

 

[madmatt30]

poor psu mate to be fair - when it was new it would be an overspecced 350w at best
with capacitor aging max output can take a nosedive over a couple of years on cheap models like that
this will determine the culprit,be warned it will push psu.cpu.gpu & temps to their limits - it will determine the problem though.

download hwifo installer (dependant on your os 32/64 bit windows)
install & run it in sensor only mode
http://www.hwinfo.com/download.php
drag down the table so you can see your cpu clock/usage & the cpu temp

download prime95
http://www.majorgeeks.com/files/details/prime95.html

again version depends on your os

download furmark & install& run (dont start a test yet)
http://www.geeks3d.com/20141110/furmark-1-15-0-released/
open prime95 extract & run,clickc just stress testing ,it will default to the blend test - click ok to start
leave it running 5 minutes or so,keep a check on your temps.
after 5 minutes click the test tab & stop.
if youhave no problems while its running the your cpu is a-ok & stable

now on furmark - untick fullscreen
set resolution 1280x720
anti aliasing off
click gpu shark - this will open a monitoring tab
click gpubenchmark ( users settings)
let it run the duration ,keep an eye on temps in gpu shark - if youre uncertain close it down.
if it runs the duration I would consider your gpu stable.

lmk how you go on with the above ,if no problems there then we need to stress both cpu/gpu at the same time,there is aknack to this though so do the above first & post back.

EDIT - just saw your speccy tab - if its reading right those voltages are all over the place.(in a bad way)
check the voltages in hwinfo before tyou run any of the above tests mate.
If theyre right you have a major problem with the psu

 

[Tomasoff]

Okay, did what you asked and I ended up with this:

1. PC didn't shut down at any point.
2. HWMonitor's data while doing the CPU stress test - http://gyazo.com/72839c58cc84be51ea57617b4d027600
3. GPU's FurMark stress test's results - http://gyazo.com/66123a65c1fd0a9db8f306a59276cb77

EDIT: By the way, didn't do both stress tests at the same time yet.

 

[madmatt30]

Okay, did what you asked and I ended up with this:

1. PC didn't shut down at any point.
2. HWMonitor's data while doing the CPU stress test - http://gyazo.com/72839c58cc84be51ea57617b4d027600
3. GPU's FurMark stress test's results - http://gyazo.com/66123a65c1fd0a9db8f306a59276cb77

EDIT: By the way, didn't do both stress tests at the same time yet.

not as simple as running both simultaneously mate is the problem,prime will take all the core usage ,which then means furmark cannot make full use of the gpu.
Im not sure if this will work on a 2 core processor with no hyper threading but try this
(you dont need prime here - you can close it down)
with furmark on main menu screen
start cpu burner
leave threads on 6 ,click start to get it up & running
right click on your windows taskbar at bottom of screen
click start task manager
goto processes tab
find cpuburner.exe
right click it
click set affinity
you should only have cpu0,cpu1
untick cpu1
click the red x to close box
find furmark.exe
right click,set affinity ,untick cpu 0
close task manager

now with cpu burner still running start benchmark users settings
this should stress both cpu & gpu 99-100%
to check this on hwinfo both core clocks core #0 & core#1 should be at current 100%
at top windows of furmark test it should show 95- 99% gpu usage





do you have the gigabyte easy tune software installed that came with your board? - this will give you relaible voltage readings

your voltages on speccy show the same as on hwinfo,if theyre correct then with a 12v load below 9v Im surprised the pc is even running,it may be that are reading voltages incorrectly - it does the same on my asus board,the gigabye software itself is the most accurate way to check

 

[Tomasoff]

Thank you for your effort mate, I really appreciate it! Today I'm going to see my father at the clinics, but in the evening I think I can manage to do those tests. And by the way, I think I have x4 cores, because I used to select all 4 for counter strike or other games.

 

[madmatt30]

You're right mate - didn't think that old i3 had hyperthreading but it would seem it does.

Well use 3 cores for cpu-burner then & use the last core for furmark.
Same as above just untick the last core for cpuburner as the info above & untick the first 3 cores for furmark.

As I said first try your own gigabyte boards monitor software to check those voltages ,if that 12v really is running at less than 9v its your main problem.
You can very likely also check the voltage lines in bios.

 

[Tomasoff]

Ok, so I did the test and PC passed it fine, CPU was about 60-62 C, and GPU ~80 C. Tried to download EasyTune, but when I started it, it showed me nothing - http://gyazo.com/e00bc600368e34ac5d82c4c0494dea23

 

[madmatt30]

In all honesty mate - your problem seems so erratic it has to be something power related IMO.
Those voltages are just plain wrong - an erratic 12v can cause bsod,freezing,shutdowns,restarts etc.
As we're aware its a fairly crap PSU at best (when it was new even) its got to be the first port of call mate.

 

[Tomasoff]

So what PSU would you recommend ?

 

[madmatt30]

Anything half decent - where do you live?

 

[Tomasoff]

Lithuania, but I can buy it from eBay.

 

[madmatt30]

Trying to find you something cheapish that will do the job.
If you can order from eBay.de this is a reasonably solid cheapish PSU that will do the job.
http://www.ebay.de/itm/400838240521?redirect=mobile

 

[Tomasoff]

Thanks dude, ordered

 

[madmatt30]

Fingers crossed mate ;-)
In all honesty you needed to replace that PSU anyway

What price did you pay?? Its gone up 15 euros since I looked??