Thanks for your guide, I just flashed both the GPU's on my GTX 690 with unlocked BIOS and have set the TDP Max to 150. Didn't really improve on anything though, the second I go past +140 (1160MHz) on the core clock it will fail during MSI Kombustor stress testing. And I think you might have misunderstood me here, I'm not claiming GPU2 in my GTX 690 to clock worse than GPU1, they are both running: +150 Power Limit +140 Core Clock (1160MHz) +500 Memory Clock (7GHz) That's through MSI Afterburner and both nVIDIAInspector and GPU-Z reports the boost clocks to be right for both GPU's. What I cant understand is how one GPU takes a trashloads of work during stress testing with MSI Kombuster's various test, while the second GPU isn't remotely close to being pushed at all. As you most likely know already the GTX 690 is basically 2x GK104 on the same PCB.. So I basically have 2x GTX 680 in one single graphics card. So I've got a GPU1 and a GPU2, but being clocked to the exact same speeds using MSI Afterburner, and they both now feature their own unlocked BIOS. What I find odd is that during various stress test this is what both GPU-Z, nVIDIAInspector and MSI Kombuster itself tend to report: GPU1: TDP: 120-140 Core Clock: 1175MHz Mem Clock: 3505MHz Load: 98-100% Temps: 75-82 GPU2: TDP: 25-35 Core Clock: 914MHz Mem Clock: 3505MHz Load: 0-35% Temps: 30-40 It's rather obvious that GPU2 is barley in use at all, and 914MHz core clock is telling me that it's boost clock of 1175MHz don't even kick in. How come? It has barley any workload to begin with, its TPD is low and its temps are way below anything that should point at throttling of any sort. Is it just bad support for SLI in the stress test making it utilise only one GPU?