PROBLEM : Can't find whether GPU is overheating or the CPU is.

[harryrupam]

I have
1. a Nvidia GT 630 1 GB DDR3 gpu (With Latest 355.60 driver & DX11 installed)
2. Corsair VS 450 PSU,
3. 3 GB ram & 80 GB HDD Sata (old)
4. Core 2 Duo E7200 @ 2.53
I've been using this card with this config. & playing games for 1.5 years without any problem but
recently (a week) sometimes when I play games (sherlock holmes/PES 15/Dead Island) screen started turning black (as if display signal is lost) & the computer crashes/hangs with a looping "zzz" sound.. :'(

I could hear noisy fan sound & I thought that CPU was making that noise (as I had 8600 GT before with the same CPU & its fan hasn't been cleaned of dust since I bought it & the GT 630 is rather "newer" compared to my CPU)
BUT I found out that my GPU was actually making that noise. I cleaned everything & played games to test it & now the problem worsened. Those games,which I played with this card at 40+ fps are now working fine at first but after 2-4 minutes, starts stuttering & ultimately hangs as the manner I mentioned above. The game uses sometimes 60-90% CPU. I noticed CPU reaches 50-60 degree while gaming but the GPU absurdly reaches 100+ degree C (with full fan speed). When Idle,GPU reaches 45-65 Degree C.

I played 3 Games (which ran fine before this problem) but all of them hangs the computer within 4-5 minutes of playing..!!! I can't play any games as within 5 minutes of every game they hang the PC. BUT without gaming,EVERY other work is just working fine. I have posted some pictures of the temperatures while idle,while gaming & right before the crash. Here is the link -

http://s946.photobucket.com/user/harryrupam/library/?so...


Can anybody help me to find out what's causing the hang? CPU or GPU? Thank you


EDIT : I have noticed that after staying idle,starting the game rapidly increases the temperature of the GPU... like from 45 to 80 Degree C within 10 seconds. And,within 5-7 seconds of killing the game with Task Manager, the Temp also decreases very dramatically... (from 100/96 degree C to 50 Degree C within 5-10 seconds!!). Is this extend of rapid heat development/dissipation even possible? the Temp almost instantaneously drops 40 degree after quiting the game... what's happening?

NB. : I have monitored all data using "Open Hardware Monitor".
I have ran the games with the cabinet open & every time of the hang,I've noticed that the CPU fan keeps spinning but the GPU fan stops spinning.

 

[LTC8K6]

The GPU fan could simply be failing and not giving accurate RPM readings. Obviously if it is stopping when the GPU temp is high, then it's a defective GPU fan.

You may need to remove the GPU heatsink and clean off the old thermal compound, and apply new compound while you are at it.

 

[LTC8K6]

You can try pointing a house fan / box fan at the open case, and see if that allows the GPU to stay cool enough to game.

 

[harryrupam]

Yeah I've ordered thermal paste & I'll apply it very soon but my question is -
IF it really was an cooling issue,how come the temperature falls so unrealistically? As if instantaneously it dissipates 50c in 2-3 seconds.

 

[dark zero]

It could be the GPU... the GPU have way shorter lifespan than the CPU

 

[LTC8K6]

You said that the GPU fan was not spinning. That is certainly a problem.

I have ran the games with the cabinet open & every time of the hang,I've noticed that the CPU fan keeps spinning but the GPU fan stops spinning.

 

[iiiankiii]

Yeah I've ordered thermal paste & I'll apply it very soon but my question is -
IF it really was an cooling issue,how come the temperature falls so unrealistically? As if instantaneously it dissipates 50c in 2-3 seconds.

yeah, the GPU/CPU can increase and decrease in temperature VERY quickly. Under load, it'll shoot up to 70-80c and drops down almost immediately to 40-50c when it goes to back idle. It is normal.

However, 100c for the GPU is way too high. It's pretty clear that your GPU is overheating. You might have clogged the heatsink/fan with dust, fan stopped working, and/or the thermal paste might be failing to the point where it's unable to create a solid contact between the GPU and the heatsink.

 

[harryrupam]

Yeah I also think that my GPU's fan/thermal paste is not working properly. Even after I cleared the dust off the GPU,it still is making noise.

The Gpu Fan spins as usual (while gaming & during idle) but at the instance the PC hangs,the GPU fans stops spinning. Just when the GPU hits 100c+ the fan spins frantically,making a good amount of noise then ultimately stops spinning & hangs the PC.

 

[Ketchup]

The MAX for that GPU is 98 C. In a well-ventilated case, it should not be hitting that, or getting too close.

So I agree, lets make sure the fan on the video card is working properly, as an overheating video card does explain your issues.

 

[KingFatty]

Straightforward issue: your GPU fan is broken, so the GPU overheats when it is used (gaming) and results in crash. When the GPU is not used, it can still function because it doesn't generate lots of heat when unused.

There is no reason to question the CPU at this point, just focus on fixing the GPU problem.

 

[Stuka87]

Do not pull your heat sink off the GPU. thermal compound does not just fail. Either it was bad from the factory, or it is fine. (Yes too much can make it cool slightly worse, but that also is a day one issue).

Replace the fan and see how things go. If you are not experienced with removing and applying heatsink compound you can make it worse than it is currently.

 

[harryrupam]

As I found that someone mentioned somewhere that Mining Malware also can cause Overheat as they put extra load on GPU... I don't know how but still I did a format & fresh reinstall of the OS ... & it's still the same... Clearly Malware wasn't the issue... it's fan/heatsink problem...
The funny thing is - I can't even let my Windows to rate itself,cause if I do,even the direct3D test for GPU (which is conducted during windows hardware rating procedure) is pushing my graphics card to 95-100 degrees and eventually hanging my PC Even Low profile games like Medal of Honor Allied assault/Sands of time are sheerly unplayable now lol

I don't know how to change the fan but I've changed paste for my CPU before. SO I think when my Thermal Paste arrives I will be able to change it
I'll give update on that when it arrives

 

[LTC8K6]

Might want to just replace the card.

http://www.newegg.com/Product/Product.aspx?Item=N82E16814133542&cm_re=gt730-_-14-133-542-_-Product

http://www.newegg.com/Product/Product.aspx?Item=N82E16814127788&cm_re=gtx750-_-14-127-788-_-Product

 

[Ketchup]

If the fan on your video card is running at full speed in 3d, I would try replacing thermal compound first.

If the fan isn't spinning, or you don't feel comfortable applying new compound, just get a new card.

 

[harryrupam]

I gave the card to local repair-shop... they tested it & verified that both the fan & the Fan Controller in the GPU's PCB is damaged.... it can't be repaired :\
Unless I have warranty,(which I don't) the card is going to the junk yard :\
I'm buying R7 250x instead of GT 740 OC in the same price range,as it performs better as most people say. I could've gone for GTX 750 but here in india the 250x costs $100 & sadly the GTX costs $140 - way overpriced.
only similarly priced card is 740 oc which I guess inferior to 250x,right? :|

 

[LTC8K6]

250X is a big improvement over the GT630.

250X vs GT740oc is about even.

 

[sm625]

I gave the card to local repair-shop... they tested it & verified that both the fan & the Fan Controller in the GPU's PCB is damaged.... it can't be repaired :\

That is bunk. All you have to do is rig a fan. Find any fan that will fit in there, and wire it directly to the +12V supply via one of the power supply's molex connectors, OR if it is a 5V fan, wire it to a fan header on the motherboard. There are plenty of options. You wont have any control over the fan speed, but that is ok. It can run all the time its no big deal. It's not worth replacing a card over, unless you want a better card.

 

[KingFatty]

Also you can use an adjustable fan controller installed in a 3.5" or 5.25" bay of the computer, so you can manually adjust the fan speed (e.g., make it fast when gaming, dial it to slow when not gaming).

 

[boozzer]

That is bunk. All you have to do is rig a fan. Find any fan that will fit in there, and wire it directly to the +12V supply via one of the power supply's molex connectors, OR if it is a 5V fan, wire it to a fan header on the motherboard. There are plenty of options. You wont have any control over the fan speed, but that is ok. It can run all the time its no big deal. It's not worth replacing a card over, unless you want a better card.

it is a gt 630, why not just buy a cheapo gpu to replace it?

 

[harryrupam]

Well, I guess bad luck never leaves my tail
I received my R7 250x,connected to my PCI x16 v1.0 & my computer won't boot... It doesn't even show anything in the screen & my machine keeps rebooting automatically.
The Box says it needs a 450W psu & a 6-pin power connector. My Psu is = Corsair VS450. I have seen in youtube & other places that many people owns this psu & run this card without any problem. To check if that card needs more power,I took that card to my brother's PC,which has Cooler Master Thunder 450W psu. which has EXACTLY the same specification as my psu. But,there my card runs flawlessly. I don't know,has my PCI-E 6pin been damaged? Or can it be a faulty PCI-E x16 slot? Or,simply my psu somehow fails to feed enough juice?

p.s. = The new GPU fan works (until the pc goes to reboot) & my old gpu, GT 630 (faulty) ,which doesn't need any 6-pin connector,also works in this slot. (though the GT 630,having the heating issue can't play any games,it can handle everything else just fine from this same PCI x16 slot)

My Specs =

1. Powercolor R7 250x
2. Corsair VS 450 PSU,
3. 3 GB ram & 80 GB HDD Sata (old)
4. Core 2 Duo E7200 @ 2.53

 

[LTC8K6]

What motherboard?

A Gen1 slot could certainly have issues with a Gen3 card.

 

[LTC8K6]

EVGA GT730 02G-P3-3733-KR

GT730 is PCIE 2.0, and should work in your 1.0 slot and have similar performance.

You want a 64bit GDDR5 384 core version. It's the fastest GT730 Version.

http://www.geforce.com/hardware/desktop-gpus/geforce-gt-730/specifications

 

[harryrupam]

My Motherboard is MSI 945GCM7
It has PCI E x16 (no version is mentioned,I guess it's v1.0)

 

[LTC8K6]

http://us.msi.com/product/mb/945GCM7F.html#hero-specification

Yes, it's the old Gen1, so that could well be the problem.

A Gen2 card should be okay.

 

[harryrupam]

http://us.msi.com/product/mb/945GCM7F.html#hero-specification

Yes, it's the old Gen1, so that could well be the problem.

A Gen2 card should be okay.

Yeah you're right. I took everything out from my cabinet,went to my brother's home & ran my new R7 250x in my brothers mobo which has PCI - E x16 v2.0 with my own PSU = Corsair VS 450. And it ran perfectly. But when I connected his PSU (Cooler Master Thunder 450w) to my mobo with my R7 250x, I faced that same rebooting problem which I faced in my home.
So, it really is my old PCI-E v1.0 causing the problem while connected to a v3.0 card. Right?
What do I have to do? A new mobo with PCI E x16 v3.0? or a mobo with v2.0 will work just fine? :|