I benched 2x6990s in 3dMark11 and here are the results:
980X @ stock (26 multiplier)
X3249 no CF
X3715 Enabled CF
980X @ multiplier 28
X3252 no CF
X5475 Enabled CF
980X @ multiplier 29
X4808 Enabled CF
went back down to multiplier 28 still get 4800 ish.
Looking at GPU-Z the main cards clock is at 550MHz. Two of my cards are near each other, so main cards air intake is effected. In all cases the 6990s are at stock clocks/volts
Conclusion:
1. 980X at stock speed holds back 2x6990s
2. 6990s if installed near each other will heat up and throttle down, which means water cooling should yield big improvements in long term performance, preventing the cards clocking lower than they start off with...
Not going to do any voltage mods to cards/cpu while on air.
I did notice earlier that i sometimes get X3000 with single card and X3300 with xfire, it now makes sense... those low score runs in the past must have been with "hot" slowed down cards.
980X @ stock (26 multiplier)
X3249 no CF
X3715 Enabled CF
980X @ multiplier 28
X3252 no CF
X5475 Enabled CF
980X @ multiplier 29
X4808 Enabled CF
went back down to multiplier 28 still get 4800 ish.
Looking at GPU-Z the main cards clock is at 550MHz. Two of my cards are near each other, so main cards air intake is effected. In all cases the 6990s are at stock clocks/volts
Conclusion:
1. 980X at stock speed holds back 2x6990s
2. 6990s if installed near each other will heat up and throttle down, which means water cooling should yield big improvements in long term performance, preventing the cards clocking lower than they start off with...
Not going to do any voltage mods to cards/cpu while on air.
I did notice earlier that i sometimes get X3000 with single card and X3300 with xfire, it now makes sense... those low score runs in the past must have been with "hot" slowed down cards.