Overheating Symptoms but not overheating!

[RzNadare]

I've been using a PowerColor R9 280x that I got from RMA less than a month ago and for the past week have been getting what I guess are GPU crashes.

This usually happens on high load when gaming: my monitor would suddenly fail to receive video input but the computer would continue to run (I can't do anything except force shutdown).

I've searched online and most responses seem to say its because of GPU overheating, but I've once tried recording the GPU temperature with MSI Afterburner and temperatures recorded up until the crash shows below overheating temperatures. Other responses suggest that the Power Supply is why, but I have not gotten these kinds of crashes UNTIL I started using the R9 280x replacement and it has enough wattage to run my computer.

I've tried stress testing to see if it would crash again using 3D Mark, FurMark, and OCCT. Of the three, 3D Mark and FurMark had no problems, but OCCT caused the no video input crash to happen within less than 15 minutes after starting it.

I have not overclocked ever since using the replacement GPU.

Any solutions would be great!

 

[LoNgMaO]

What kind of connetivity you are using. example D-sub, Dvi or HDMI.

 

[RzNadare]

What kind of connetivity you using. example D-sub, Dvi or HDMI.

I am using a HDMI cable, however I do not think the cable is the issue, as this problem only showed up after RMA'ing my GPU.

 

[royfrosty]

Most prolly your psu.

 

[LoNgMaO]

Try do a gpu ram test. The above test you did is only to stress test the gpu.

 

[RzNadare]

Most prolly your psu.

Thank you for contributing, but I am already aware that the PSU could be an issue. If the PSU was the case, I would've had this issue from before I RMA'ed my GPU. In addition, all of my PC components are very new (Less than 2 months ago) so it could not be due to wear and tear.

 

[RzNadare]

Try do a gpu ram test. The above test you did is only to stress test the gpu.

Is there a reliable GPU ram test you can suggest for me? I've Googled for GPU memory test softwares and found these, but I don't know if they are reliable or kept up to date:

1. CUDA GPU memtest
2. Video Memory Stress Test
3. MemtestCL 1.00

 

[LoNgMaO]

Sorry, I don't have one to recommend. You have to try an error. Thanks

 

[LoNgMaO]

I have study your computer specification. Could you run your game using your IGP see if the fault does come from any other part of your system.

Also I always like to ask ppl to clean their ram and gpu card gold finger cause something oxidation will cause errors and hang.

 

[RzNadare]

I have study your computer specification. Could you run your game using your IGP see if the fault does come from any other part of your system.

Also I always like to ask ppl to clean their ram and gpu card gold finger cause something oxidation will cause errors and hang.

By IGP do you mean the Internal Graphics processor? Even if I do though, it won't necessarily occur at a particular point in time because it just happens unexpectedly. I will have some games where it won't crash at all, and other times it has happened when I wasn't playing (Alt tabbed out).

I assume the "gold finger" you're talking about is the bit that slots into the motherboard? I've never heard of oxidation on RAM and GPU but wouldn't that take a long time before it becomes an issue? As far as I know, the PC components should be fine because they were all newly bought less than three months ago and was assembled by the staff at the same store I bought it from.

 

[hlots123]

but OCCT caused the no video input crash to happen within less than 15 minutes after starting it.

how abt reducing gpu core and/or mem clocks e.g. by -50. or upping voltage a tinnny bit? by chance, if no more occt "no video input", at least can know is gpu problem...

p/s does this card show vrm temps?

 

[RzNadare]

how abt reducing gpu core and/or mem clocks e.g. by -50. or upping voltage a tinnny bit? by chance, if no more occt "no video input", at least can know is gpu problem...

p/s does this card show vrm temps?

I've just tried underclocking by 50 and 100 on both core and memory clocks. I do still get "no video input", but the tests had lasted longer than base clock speeds. But if one has to underclock by over 100 from base clock, isn't that already a sign that the GPU is faulty then?

I did not try upping voltage because I have no experience with it and I'm afraid of causing more damage recklessly if I do.

I don't think my card shows vrm temperatures. I don't know what to look for though (I have HWmonitor, MSI Afterburner, and OCCT but none displays vrm)...

 

[hlots123]

But if one has to underclock by over 100 from base clock, isn't that already a sign that the GPU is faulty then?

yah, suggestn was to try to see if able to ascertain gpu or psu fault etc. since u wrote below... but hoping to cfm gpu, so u can re- rma again.

I've searched online and most responses seem to say its because of GPU overheating, but I've once tried recording the GPU temperature with MSI Afterburner and temperatures recorded up until the crash shows below overheating temperatures. Other responses suggest that the Power Supply is why, but I have not gotten these kinds of crashes UNTIL I started using the R9 280x replacement and it has enough wattage to run my computer.

 

[FinalTidus]

I think could be PSU fault...

 

[RzNadare]

Ok, so I had just finished several one hour tests on OCCT changing the Core and Memory speeds by using MSI Afterburner:

Base Core & Memory speed (1030, 1500): "No Video Input" within less than 5 minutes.

Core & Memory speed -200 (830, 1300): Did not crash into "No Video Input".

Core -200 & Base Memory speed (830, 1500): Did not crash into "No Video Input".

Core & Memory speed -100 (930, 1400): "No Video Input" within less than 20 minutes.

Core -100 & Base Memory speed (930, 1500): Did not crash into "No Video Input".

So based on my tests, I'm fully convinced that the GPU is the cause of the "No Video Input" crashes, but what I found odd was that underclocking both the core and memory clock speed by 100 crashed the GPU, while underclocking ONLY the core clock speed by 100 allowed the OCCT test to be completed.

Whats disturbing though, is that the test at base Core and Memory speeds appears to crash into "No Video Input" even faster than before. This was also the first test done immediately after turning on the computer, so the shorter crash time is not due to cumulative testing.