Quote from: Bufius on February 08, 2014, 10:34:51 PM

A word from my side about crashing miner and hardware

Since some days i have a VTX3D R9 290 X-Edition card and there were and are issues with miner crashes and driver recoveries at stock clocks.

BUT: i know why this happens!

1) after i got the card and put it into my win 7 PC this thing was very very hot. even at 700 MHz i had 93°C with Intensity 14. Higher intensity: crash
-> issue fixed: the original applied heat compound was crap. air gaps inside and some of the chip was not covered. Now with new heat compound it is much better than before.

2) after this heat compound thing i tried higher clock rates.
max stock 975 MHz and Intensity 15: crash after a few seconds. With Intensity 14 crash after <1 minute.
With 900 MHz: really hot but working, even with Intensity 20.
Test with furmark: with 975 MHz and max settings: no problems, works fine for > 5 minutes (just a short test performed)

Then i started HWInfo and checked the Voltages: with Furmark, the GPU voltage is at min somewhere at 0.98xV @ 975MHz.
With vertminer voltage is at 0.94xV @ 900MHz. Also the temps are not as high with furmark @ stock than with vertminer and lower clocks.

-> my conclusion:
for GPU testing purposes, furmark is a toy. If you want to perform real testing, use vertminer. The opencl kicks the GPU to the max and so everything else (regulators, memory) is used extensive.
And if the cards stock voltages are good enough to play games, this does not mean you can use the card with the same clocks for mining.

--> final word
If you encounter crashes, lower your clocks or increase the voltages! And a good cooling is a must!
The latest opencl kernel for the vertminer is even more optimized than the existing cgminer scrypt kernel. You can try to use the latest vertminer kernel also with scrypt cgminer (but needs some changes to work).
I am sure you will see also some sudden crashes when your card is at its limit...

Bufius,

Are you sure the card is OK? I've got Sapphire R9 290s (non-X), the standard reference design (with the hairdryer-like noisy fan).
With the engine @ 947 and mem @ 1500 and Intensity 18 my cards get 76C. Fan runs at 75%.

Higher intensities seem to crash the card at some point, as does a higher clock. Just FYI. I find 93C at 700 wit I14 a bit suspicious.
Edit: Ok, temp was before you fixed the paste problem, but still the crashing is odd. I *can* run the cards at 1000 clock, but they crash after hours/days then. At 85C btw.