GTX 460 Display issues during GPU intensive operations.

[DPJenkins]

Hello all, first time post - seeking some opinions and ideally solutions to my current issues.

I am having a problem ONLY during GPU heavy operations, my display is cutting out as if there hdmi cable has been pulled out "no input detected". I've got the latest drivers. I downloaded a video stress test (FurMark) and it crashes every time after only around 15 seconds. My computer remains on, with the fans blowing pretty hard - and I have to hold down the power button to turn it off. The good thing about the stress test is that it displays the fan power and GPU temperature, which remains below 50, usually 30-45.

I thought it may have been my PSU, so bought a good reputation antec 650 true power PSU, and after installing it today full of hope... it remains the same.

Custom built in Oct 2010, but I've been busy and avoided games apart from World of Warcraft (which it doesn't crash on) so this problems only recently been my focus. The spec is:
EVGA GTX460.
Gigabute H55M-UD2H mobo.
Inter Core i5 CPU 760 @ 2.80GHz, 4 cores.
4gig (2 sticks) corsair RAM.
Antec 650W truepower PSU (brand new).

I've written to both the company I bought it from, and EVGA to enquire about a possible warranty claim.

ANY advice is welcome, and I would be highly grateful to anyone who takes their time to try and help me understand and possibly fix my issue. Thank you.

 

[bucknutty]

Many of the EVGA GTX460s were overheating for no apparent reason. The temp goes from like 70 up to over 100 in like 1 minute and the PC shuts down. My Buddy has one and it had issues during a lan party at my house. I had to lend him one of my GTX 470s for the rest of the weekend. Any way this was in February, his card was 6 months old at the time. Evga replaced it in like 4 days and he has been fine ever since. You cant beat EVGA and their life time warrenty.

 

[bucknutty]

Install EVGA Precision or MSI afterburner and watch the onscreen display when gaming. Anything in the 90s above normal.

 

[ilysaml]

have you tried any other input cable such as DVI ? or you only use HDMI?
how long have been this issue happening ?

 

[DPJenkins]

Thanks for the quick replies. Well I'm almost convinced it's not a heat issue because during the stress test it deplays the temperature which stays approx 40, which is really far from dangerous.

Regards to how often this has happened: Around 4 months ago it happened during some game (can't recall which), and so I reformatted in hope to fix it (PC was due one anyway). It didn't happen then on Warcraft, but as soon as I tried deus ex (around 2 weeks ago) it's started happening. It now happens on Counterstrike source (less often), and has happened just once on Warcraft.

Regarding warranty, I was unfortunately blissfully unaware of the warranty stipulations. I registered it 30 exactly after my Due delivery date, and so I'm hoping they accept that. Also, the sticker with serial number etc on the graphics card has peeled off over time, and was hanging on effectively when I inspected it a few weeks back. so I've carefully taken it off for now - the sticker doesn't say "DO NOT REMOVE" but then their website warranty information says it can't be removed. I've written a message to them asking their opinion on 1) the issue, and 2) my warranty validity. Finally, it says on EVGA website if I didn't buy it through their website, I should contact my supplier - which was "scan". So I've contacted them also now, awaiting reply form both.

I will download precision incase this graphics stress test temp monitor isn't accurate. Thanks SO much again for replies.

 

[DPJenkins]

I tried the standard video connection instead of HDMI and it still crashed on the video stress test. To my understanding - the video stress test shouldn't crash a computer unless there is a heating problem or a psu problem. Am I right?

 

[benski]

40 degrees during a stress test is totally abnormal and makes me think either the test wasn't run properly or you card is stuck at idle clock speeds and voltage. Download MSI afterburner and kombustor, run kombustor while afterburner is running and see what the core and mem clocks are as well as the voltage while the test is running. If it still shows 40 degrees I suspect the card is defective.

 

[DPJenkins]

Alright, I downloaded afterburner v2.1.0 and Kombustor v2.0.0 was with it. I first opened afterburner, and my computer was idling at around 42C.
The values displayed on the afterburner application for my GTX 460 is:
Core Voltage (mV): blank
Core Clock (MHz) 720
Shader Clock (MHz) 1440
Memory Clock (MHz) 1800
Fan Speed (%) Auto 30.

I decided to run the KMARK BENCHMARK Preset: Normal (1280x720 fullscreen). It was fullscreen so I wasn't able to monitor it in afterburner, though I looked at the online display during the test.
GPU Temp: 98
Core Clock: 720
Mem clock: 1800
Shader clock: 1440
VDDC: 0.950V
GPU usage: 98.0
Mem usage: 488.1MB

It didn't crash, which was nice. I'll post this and try a more intense test now.

 

[DPJenkins]

Also, here are the details from the MSI Kombuster Score screen, under Benchmark Data.
kmark normal preset (Open GL 4).
Graphics score: 1904 points (36 fps)
PhysX score: 2529 points (67 fps, 141 sps)
Combined Score: 1667 points (10 FPS, 86 SPS)
PhysX version 9100514 (PhysX GPU)
Res: 1280 x720(fullscreen)
Antialiasing: 4x MSAA - PostFX: ON
FPS: min=10, max=77 avg=31
Max GPU temp: 99 C

 

[DPJenkins]

I just ran the "Burn-in test" on default settings, and after 40 seconds of running it crashed in the way described in my original post. CPU temp got to stuck at around 100C.

 

[DPJenkins]

I played some games quickly then with the on-screen display to monitor the GPU temp and fan. Played world of warcraft, temp seemed 80C, fan at quite a low speed. When I played counterstrike source, the temperature was constantly above 90 degrees, and fan around 75%. When temperature got to 98 or 99, fan would switch to 100%, it'd go back down to low 90 degrees for a little. This cycle loosely repeated on a frequent basis. I didn't see the temp reach 100.

I check the 460 spec, and it says the max temp is 104C. Does this mean it would break the GPU if it got that high? In which case, isn't my card running pretty damn hot even though it isn't overclocked at all and running counterstrike source, which is a relatively old game? I'll try it in Deus Ex tomorrow (it always crashes on that) and see if the temp peaking has a relation to the crash.

 

[benski]

Thats WAY too hot. It won't fail right away but the heating up and cooling will cause it to fail pretty quickly. I have a GTX 460 in my HTPC and it never gets over 75 degrees running furmark or kombustor and never comes close to 70 in normal use. Return the card, it should not be running anywhere close to that hot even with the default fan profile.

 

[DPJenkins]

Thank you for your input, this was my first build and so unfortunately I was pretty naive about what temperatures these things run at.

 

[DPJenkins]

Just an update on the situation, EVGA agreed that temperature is way too hot, and I've had a RMA given to me, and sent back my card today. £15 shipping unfortunately, but if it fixed my situation over all it's not so bad I guess.

I've found EVGA quick to respond and helpful, as I have with the feedback given on this forum. Thank you and I'll let you know when I receive my card.

 

[bucknutty]

Dude i called it. I am always right sometimes!

 

[DPJenkins]

Hey gents, just a quick message to give you an update. Received my RMA GTX 460 back today from EVGA, and first thing I did was run the same stress test I did before.

I am pleased to say the results were great. The temp didn't go above 58, and slowly ramped up from 31 to that over the 2 minutes. Nothing like the 99C in 5 seconds I had before.

Here are more details results for those interested:
kmark normal preset (Open GL 4).
Graphics score: 2674 points (49 fps)
PhysX score: 2673 points (77 fps, 147 sps)
Combined Score: 2135 points (17 FPS, 102 SPS)
PhysX version 9100514 (PhysX GPU)
Res: 1280 x720(fullscreen)
Antialiasing: 4x MSAA - PostFX: ON
FPS: min=17, max=91 avg=41
Max GPU temp: 58 C

Not only is the temperature normal, but it seems to be performing significantly better than before which pleases me.

I'd like to close through thanking all who tried to help me during my difficulties. You gave me the confidence and advice to resolve this, so thank you.

 

[benski]

Good to hear. Good luck, with those temps you might consider trying out some GPU overclocking software. You can get alot of extra performance out of a 460 without causing much more heat.

 

[nforce4max]

Many of the EVGA GTX460s were overheating for no apparent reason. The temp goes from like 70 up to over 100 in like 1 minute and the PC shuts down. My Buddy has one and it had issues during a lan party at my house. I had to lend him one of my GTX 470s for the rest of the weekend. Any way this was in February, his card was 6 months old at the time. Evga replaced it in like 4 days and he has been fine ever since. You cant beat EVGA and their life time warrenty.

There is a reason why they overheat, take a look for your self and take notice that there is nothing at all to cool the power vrm on almost all gtx 460 models except for a small few exceptions. The power vrm is rated to run around 115-125c but not indefinitely so they gradually degrade over time till point of failure. The mosfets in the power vrm phases is a typical analog flying wire design (bonding and ic type) so during thermal changes during normal use and off gradually weaken the bonds till they allow for the wire to break away from the pad on the ic thus killing the circuit until high enough temp is reached when the ic and wire has expanded slightly from heat to complete the circuit but every time the temps reach higher and higher than when it first broke it pushes the wire further away from the ic so when the card is cool the mosfet won't be functional. The other problems are some times the wire or ic suffers a short.

The vram (memory) is also very poorly cooled on almost all models as well except for a little air flow from the cooler but usually not enough to keep temps down but GDDR5 runs cooler than GDDR3 and 4. Some cards do suffer some failures here but they are easy to spot when they come up.

The core sometimes comes with a few defects such as a bad IHS that isn't flat or poor application of thermal compound under the ihs leading to higher than normal temps even for that model. Bad bonding under the gpu die or substrate often cause problems when present.