If it downclocks, it's either because your GPU hits 70C° or because the power target is too low for your OC.
Valley 1.0 isn't reporting the proper boost that's all. List my boost at like 1354mHz when it is actually 1267mHz.
Oh, well.. At least it does report my clocks correctly ^^ I'll try to remember that, thanks (I did read this before now that you mention it). Anyway, here's an update on my score with some tweaking : - System RAM sub-timings are much tighter (+1.5 FPS from that only - maybe less hiccups.. Not sure). - 13 MHz less on the GPUs boost clock (so 1267 MHz instead of 1280). This last part is actually funny, because 1280 MHz never failed on me and I'm using this clock for 6 months or so now.. I ran multiple tests and I'm losing a few FPS running 1280 vs 1267 - ON THIS BENCHMARK ONLY. Pushing to 1293 or 1306 gives a boost again, but I can't play BF3 for an hour without a crash, so.. Maybe a sign of slight instability @ 1280 for my cards (or PCI-E 2.0 is acting up - unlikely). Also, I did update drivers to 314.07, but I'm not sure of any gain from that on this particular benchmark. @ TK - This benchmark is very strange. It seems CPU and/or memory bandwidth/latency affects the score to a very high extent in fact.. I'll try @ 5.3 or 5.4 GHz to see if any substantial gain is visible, but +1.5 FPS from sub-timings tweaking looks ridiculous to me..
Here's mine: Weird how I got lower perfomance than a GTX 580 ??? this is 1050/1500 stock voltage EDIT: found the issue. I had the AA setting in catalyst CC to improve quality so it was 8Qx AA rendered. After disabling it to standard application settings I got this result:
Dude, that score is plain wrong for that setup... 3x 6970s can crack around 3-4K scores ! Check your setup/GPU use !
CPU: i7-3820 @3.6ghz Drivers: 314.09 Extreme Preset @1080p FPS: 81.1 Score: 3394 Min FPS: 31.1 Max FPS: 153.1