Added washers to the heatsink screws to increase pressure on the GPU. Seems to have done the trick for clearing up the last of the rendering errors that would pop up randomly at about 60-62c after prolonged gaming. No problems up to 75c so far... though that doesnt speak well for the Asus heatsink.
With some ducting work the heatsink improves by about 5c+. All the air gaps around it are pretty bad for performance in stock form.
What GPU Tweak looks like before a crash:
https://dl.dropboxusercontent.com/u/...utweakfull.png
What GPU Tweak looks like after a crash:
https://dl.dropboxusercontent.com/u/...k2350crash.jpg
All the sensors you notice as missing between the two shots will show up as 0 or 100/110/etc while crashed before restarting 2.3.5.0. After restarting they go missing.
Second card has died.
It actually died about 2 weeks ago, but I didnt have proper hardware handy for validating that it was actually dead so it sat for a while before I could test.
Also one of my GTX 670's died.
Thoroughly disgusted.
EDIT
Do you have a "bad" card? Post your info here please.
My buddy just had a 7970 Direct CU II that died. The VRMs tended to get really hot - by that I mean +110C hot. I think the cooler was loose - only his version seems to have been with the VRMs. Core temps were normal, but memory and VRMs got really hot. Not sure what is going on, but stay away from the Asus Direct CU II cards this gen it seems. My buddy has RMA'ed his, although I'm not sure how that will go.
@STEvil, have you tried to RMA?
Also, somewhat unrelated, but over at OCN, there's been an effort to try to flash the Matrix BIOS into the Direct CU II one:
http://www.overclock.net/t/1390275/w...o-thread/0_100
I RMA'd the first card, this is the second to die.. I dont want a third and i'm trying to work out who at Asus to contact to get something actually done since none of the other reps seem to be able to do anything.
I think the Matrix 7970 was EOL'ed anyways. Often when an RMA like this happens, your best bet is to see what they have in stock. You can try to contact them and see. Also, try to ask for the reference model. I know that some manufacturers will allow you to "step" down to a reference 7970. Also, if they offer a Direct CU II, don't take - get the reference model. I know the reference model is considered a "downgrade" in a sense, but at least it doesn't have these issues.
not much talking about RMA but 2 dead cards reported here...
mine died after first ln2 session and i took really good care of it :<. 1700 core card
@STEvil
Another thing you could ask Asus is if they have a 770 in RMA stock. They might.
Also, if you do get a reference board, the other benefit is that you can use other software like MSI Afterburner to OC.
Not touching 770 with a 10 foot pole until we see if the VRM's are sturdy, and even then i'll probably sit on my GTX670 that hasnt died (have one that did die) until the next round of cards come out.
SoF - Random death or due to some abuse?
Vivi - Not looking for cards that died from abuse