Yes I had Gigabyte 290 OC Windforce model with the 3 fans and Elpida ram.
Had a reallly bad time of it. Trying to diagnose the fault was hard. I replaced the PSU to give me more power (from 500W to 1100W on one 12V 90A rail). Also unplugged most of the ram, case fans, peripherals etc.
There were a few possible other sources that totally weren't the issue, such as slightly bulging caps over my motherboard memory (it's been like that for years tho). So I didn't discount anything else from being the fault, such as the motherboard being 7 years old and unable to carry enough charge over the PCI-E slot etc. However previously i've had the 7950 plugged in.
The issue continued after PSU upgrade.
1.) Flickering lines on bootup (I use DVI) once the graphics driver is initialised.
2.) Large black blocks (indicates GPU memory issue) on windows and driver would stop start.
3.) Card won't run 3DMark 2012 (or is it 2013) Fire storm with powertune -40%, core -40% and if fan is on auto ("Uber" switched turned on). HOWEVER I could get it to run with fan on max.
4.) Card won't run 3DMark 2012 Fire storm if fan is on auto ("Uber" switched turned on). HOWEVER I could get it to run with fan on max. Card Appeared to get much lower score than my 7950. (2000 versus 4000 IIRC)
3D mark appears to stop at the section where it's filling up GPU memory etc.
5.) Powering off and on the machine after it got hot resulted in "Hypertransport flood" on bootup windows (it could be misdiagnosed as a driver bug, but this is not the case after trying it for a long time) would crash and reset PC.
Issue appears to get worse with graphics card temperature.
6.) Took it back to company I purchased it from. After some time running the card and they rebooted, it said a PSU surge was detected (they did not receive the hypertransport flood message.. could be different variants of the same thing where newer motherboards have better diagnostics on the hypertransport bus) and rebooting caused issue 5.)
Didn't play around much with memory clocks but did notice that flickering continued regardless of memory clocks
Have tried catalyst 13.12, 14.1 beta (latest one) and 14.2 beta.
Very much doubt it's a driver issue because of 5.) and 6.)
Therefore it could be VRM issue or memory heat issue or both.
I don't have a great understanding of graphic card design, however it sounds like a Voltage Regulation (heat related?!) issue as increasing the intensity in a mining software slowly was able to reach really high amounts. Only after I rebooted did the machine stop.
Not buying AMD again, at least test the cards before sending them out.
02:19 AM by