Video Output Fails on HIS 7950 IceQ Turbo During Load, Resulting in Unresponsive Computer

Gh0stf0rce

Reputable
Apr 25, 2015
4
0
4,510
Hello Everyone,

To preface, this is my build:
Intel Pentium g3258 @ 3.20 GHz (stock speed)
Stock Intel Cooler
Corsair Vengance DDR3 1x4GB Downclocked to 1333mhz
MSI H81M-P33
1 TB WD Hardrive
HIS IceQ 7950 Boost (refurbished)
500B (500W 80+ Bronze Certified) EVGA PSU
Corsair Spec-01 Case

Acer 1080p IPS panel connected to the computer with a DVI-cable

Now, onto the issue: Whenever I load any game, even one as non-graphically intense as Hearthstone, before the game loading screens finish the GPU fan suddenly goes to 100%. It makes a noise similar to that of a revving car. When it finally seems to reach peak fan speed, video output fails and there is no source detected on the monitor. After video output fails, the GPU fan goes back to ~25% load, and the computer refuses to turn off or accept input unless the power button is held to do a force shut-down. The game I have been testing the issue on is CSGO.

Also, this is a new computer assembled yesterday. It was playing a CSGO bot match fine last night. When I moved it to my room, the issues began. Therefore, one of the possible causes of the issue could be that the GPU bus was tweaked during transport, but this seems unlikely as the problem only seems to occur under load. One possible explanation is that I had been playing the game on very low settings before I got the computer, and due to steam cloud this carried over. When I turned the settings up, I didn't immediately have it crash, but the next time I tried to boot it the settings forced the GPU to go under load, and therefore make it start to crash. However, even something as graphically light as Hearthstone doesn't work anymore, so I am at a loss as to how I played that CSGO bot match.

Troubleshooting Steps Taken:
Windows Update
Checking that all power cables are properly plugged in
Re-seated the GPU
Re-installing of AMD Catalyst drivers twice. The first time followed the AMD support procedure, and the second time I used Display Driver Uninstaller to uninstall, and then I reinstalled both times using the software AMD provides
Decreasing clock speed and power consumption in AMD Catalyst control center
Temperature logs from GPU-Z demonstrate that temperature remains constant and that it isn't the cause of the shutdown
Interestingly, Rendering test by GPU-Z to check actual bus speed proceeds fine, even when card goes to max fan RPM which seems to indicate heavy load.
BiOS flash for motherboard
Manipulation of PCI-e 2.0 x16 bus latency to higher levels as it is a single GPU system
Unigine crashes when the card again spins to max RPM and goes under heavy load, ~ 3-4 seconds into the test.
Settings for Unigine:
DirectX11 – Medium Quality - AA x8 – Everything else disabled

As may be obvious, none of these steps alleviated the problem. Interestingly enough, the first time clock speed and power draw were decreased, CSGO got a little farther in loading than usual. However, attempts to replicate this result haven't met with success so I will put it down to chance.

Currently, my other two trains of thought are that either the PSU cannot supply the power the card is trying to draw, or that I received a faulty card. There is a history in Newegg reviews of this problem being an issue 1 month – 1 year down in the life of this specific card, and as the card is refurbished its age could certainly be that old. As for the PSU power issue, it has a 40A 12V rail which outputs 480W for its dual 6-pin PCI-e connectors. The card under max draw according to a Tom's Hardware Review pulls 280W at peak loads, and the total system wattage shouldn't exceed 400W based on an online PSU calculator. However, the amperage may be wrong, or the calculator could be wrong, I could be misunderstanding the wattage, etc. As such, I can't rule out it being a power issue. A final shot in the dark would be the motherboard is for some reason incompatible with the graphics card.

I received the refurbished card from GPUshack, who seem to be quite reputable and responsible, so I want to test all other avenues before I request a return from them.

Any help would be appreciated.
 

Stardust342

Distinguished
Aug 12, 2012
258
0
18,860
Have you tried this graphics card in a different rig? If it's possible you should try it.
And the temps you were recording, were they the gpu core temp? Make sure it wasn't the VRM or memory temperatures.
 

Gh0stf0rce

Reputable
Apr 25, 2015
4
0
4,510


So I checked the temps using Speccy on the GPU core temp, and it records the GPU going >100 C before it has the black screen. I must have been reading the GPU-Z readout wrong. This is clearly the problem.... do you know what usually causes this? The airflow in the case is decent, (though admittedly its one 120mm fan pulling in through the front and one 120 mm fan pushing out the back)and the IceQ cooler has the proprietary back-plate exhaust anyhow so it seems like case temperature isn't the cause. The air being pushed out is cool to feel.

However, Speccy has the motherboard temp at 50 C, I don't know if this is a reliable case temperature indicator or not.

Finally, I don't have another rig to test this in unfortunately, as I have the only desktop in my house. I'll talk to my friends, but I'm not sure their rigs even have a psu with the required 2x 6pin PCI-e plug