Hi all,
History (TL;DR: you can skip to the next paragraph)
I originally bought my (first) GV-N285OC-2GI approximately a year ago. This worked fine until I attempted to use it in a mixed-GPU setup, after which the graphics drivers repeatedly crashed (~10 times per startup) even with a fresh OS install, and would then remain stable until reset. I applied for an RMA and received a replacement over a month later with exactly the same issue. I put up with this for almost a full year until I upgraded my machine and was able to test the card with a new motherboard, CPU, PSU and ram.. same issue. So at this stage I was fairly certain I had been given a dud card, and applied for a further RMA. The replacement arrived a week ago, and im happy to say that the driver crashes have gone.
Current issueA couple of nights ago I was playing Amnesia with the GPU at approximately 85C, which seems absolutely fine to me. I paused the game and came back a few minutes later to find it running at around 110C. This did not seem right at all, and I noticed some loud intermittent buzzing noises. Because of this, I took the side panel off my case and placed my hand near the GPU fan to test for airflow, no airflow at all, I was able to touch the fan itself and attempted to spin it up (as the momentum can help in cases where insufficient voltage is supplied), it didnt work. Every so often a loud whoosh could be heard as the GPU fan spun into action (at very high RPM) for roughly half a second. So, I rushed to hard-off the machine and waited for the LEDs on the motherboard to go dark, and started the machine back up while testing for airflow. The GPU fans went straight to work cooling down the card, which was still very hot indeed, and they were once again at very high RPM, only slowing a while after windows had finished loading.
I called technical support, and they said it might be because of the game I was playing, which should not be the case. I have tried Furmark to stress the card and have been unable to replicate what happened. I like to run a fairly tight ship so I do run prime95 and memtest86+ every few months to ensure my systems are stable, so the last thing I could think of is either the fan itself, or the fan controller on the GPU might be at fault. with Rivatuner I have noticed that the sensor on the GPU can return some rather erratic readings at times, claiming fan speeds have dropped to <100rpm from the usual 1460rpm, and in some cases sharp jumps in the ambient temperature (e.g. a fairly constant 45C right now also includes one reading of -79C and two greater than 90C, which are obviously entirely incorrect). These outliers sometimes correlate with changes to the "sensor supply voltage" for the card, but I do not trust this value either, as it is no doubt also reported by the same sensor providing the other erroneous values.
So it would seem to me that the fan might have cut out because the sensor readings were incorrectly reported. I would love to re-flash the VGA BIOS, but Gigabyte do not have a copy of the original BIOS on their website and Gigabyte US (UK was already closed when I tried to call today) refused to provide me with a copy of the original. I dont really want to RMA the card since this is my second replacement already, and it seems lucky that I was able to finally get one working. I wouldnt mind a replacement with a different model, but this card is intended for use with CUDA, where the 2GB GDDR comes in very handy.. not many Gigabyte cards have >1GB these days. Another alternative would be for me to fit an aftermarket cooler, but I have been told by Gigabyte UK that this would void my warranty .. despite the fact it would be vastly more reliable to do so with the fan connected to either a molex, or PWM via the motherboard.
Any advice/comments welcome. In the meantime I will continue playing games (for science!
) and trying my best to reproduce the problem. It might have just been a once-off fluke, but it screams instability. I cant very well set the machine crunching numbers and leave the house if the GPU fan might suddenly stop and cause a risk to my hardware or other property.
Edit: following the FAQ, details of my system are as follows:
Motherboard: GA-MA790FXT-UD5P, BIOS ver. F7, Graphics card: GV-N285OC-2GI, CPU: AMD Phenom II X4 955, RAM: 4x2GB OCZ3G16002G, OS: Win7 64bit, PSU: PCP&C 750W
