Hey guys. Here are some 16x/16x vs 16x/4x SLI comparison numbers
My two cards are sandwiched in 16x/16x mode due to the placement of the pci-e slots. Did some testing by moving the second card down to the 4x slot. Doing so will greatly improve my bad temps, but how would it affect my performance? Results below
Current clocks (750/1500/1850). 75% fan
Vantage (1920x1200 performance):
16x/16x - 32698 78C/61C (top card, bottom card)
16x/4x - 30973 63C/61C
Result - 5.3% performance loss. 15C drop in temps for the top card!
Just Cause 2 - Concrete Jungle (1920x1200 max settings, 4x AA, 16x AF)
16x/16x - 72.35 fps avg 74C/59C
16x/4x - 68.28 fps avg 59C/58C
Result - 5.7% performance loss. 15C drop in temps for the top card
Mafia 2 (1920x1200, max settings, physx high)
16x/16x - 51.2 fps avg 70C/57C
16x/4x (physx set to 16x card) - 51.5 fps avg 56/55C
Result - Same fps as 16x/16x. I guess since physx is a bottleneck here, the pci-e bandwidth didn't come into play. Huge drops in temps like the other tests
Dirt 2 (1920x1200, max settings, 4x AA, 16x AF)
16x/16x - 134.8 fps avg 74C/59C
16x/4x -132.1 fps avg 61/60C
Result - 2% performance loss. Not really much of a change here. Same type of temperature decrease.
Overall not too bad at all. IMO the drops in temps are worth the slight loss in performance. The performance can easily be made up by the extra overclocking headroom gained.
My two cards are sandwiched in 16x/16x mode due to the placement of the pci-e slots. Did some testing by moving the second card down to the 4x slot. Doing so will greatly improve my bad temps, but how would it affect my performance? Results below

Current clocks (750/1500/1850). 75% fan
Vantage (1920x1200 performance):
16x/16x - 32698 78C/61C (top card, bottom card)
16x/4x - 30973 63C/61C
Result - 5.3% performance loss. 15C drop in temps for the top card!
Just Cause 2 - Concrete Jungle (1920x1200 max settings, 4x AA, 16x AF)
16x/16x - 72.35 fps avg 74C/59C
16x/4x - 68.28 fps avg 59C/58C
Result - 5.7% performance loss. 15C drop in temps for the top card
Mafia 2 (1920x1200, max settings, physx high)
16x/16x - 51.2 fps avg 70C/57C
16x/4x (physx set to 16x card) - 51.5 fps avg 56/55C
Result - Same fps as 16x/16x. I guess since physx is a bottleneck here, the pci-e bandwidth didn't come into play. Huge drops in temps like the other tests
Dirt 2 (1920x1200, max settings, 4x AA, 16x AF)
16x/16x - 134.8 fps avg 74C/59C
16x/4x -132.1 fps avg 61/60C
Result - 2% performance loss. Not really much of a change here. Same type of temperature decrease.
Overall not too bad at all. IMO the drops in temps are worth the slight loss in performance. The performance can easily be made up by the extra overclocking headroom gained.














