k, well
I have found through trial and error (and help from ckolivas) that my 7970s run below stock voltages even though cgminer is still reporting stock.
How do I know this?
plugging system into a wattmeter and comparing power draw at different clocks
Finding that if I do not put --gpu-vddc <number> in my .bat cmd line (ie stock voltage) the cards overheat and crash within 10 seconds
I have also an issue with one card running consistently hot (thanks no advice needed, the supplier doesnt care , told them I'd see em when it fails)
it took quite a while and I'm not sure if its just a problem with the sensor, but with this command line I get 702Mhash per card
cgminer.exe -o
http://eu.ozco.in:8332 -u graet.20 -p p -o
http://us.ozco.in:8332 -u graet.20 -p p -I 9 -k poclbm --auto-fan --gpu-engine 1100-1180 --auto-gpu --gpu-vddc 0.87 --gpu-memdiff -150 --temp-target 75,95 --temp-overheat 95,115
without those temp settings the card was being throttled by cgminer
I found system with 2 card enabled:
base measurement
97W idle
652W hashing with only --auto-fan --gpu-engine 1100-1180
my command line above
102W idle
620W
been running stable at 702Mhash/card for 5 days (power outage killed my uptime
)
I have been told this is impossible - meh, I believe my eyes
cards GDDR5 Overclocked HDMI Radeon GV-R797OC-3GD 7970
I don't do ref cards - 3fan gigabytes are my fav - hash well, run cool and most importantly quiet