I just downloaded and watched the video PCPer is referencing as proof in their editorial.
My issue is this.First half of the video is at full speed,second half at 50%. I saw screen anomalies in ALL samples,single AMD GPU,single Nvidia GPU,crossfire and SLi. They occur in different areas and at different times.They don't tell you which one is dual card and single card.Only thing clear is one side seems to be faster than the other at some times,but it's not consistent,at all.Furthermore,there is no quantifying data what-so-ever. If I post a claim anywhere on the internet,I back it up with some form of proof.Looking at this video I have no idea if crossfire was enabled or not.I have no idea if one card was throttling due to overheating or not. I have no idea if there is a CPU bottleneck when crossfire or SLi is enabled.I have no idea what RAM and what speed was used,because you know GPU's DO NOT work by themselves. How hard would it have been to open GPU-z and HWMonitor to show it on the screen? This is what I have a problem with.
If I claim anything on this forum I back it up to the best of my ability.That's a pretty standard practice in the PC hardware world,I do believe.There doesn't seem to be any attempt to quantify anything other than accepting your word that what PCPer say's is true.
Even all the frame pacing stuff early this year is extremely subjective,based on your proprietary hardware.I'm not saying there is no issue,I'm saying your method of quantifying these issues is pretty suspect,and not very scientifically reliable. Way too many factors involved,from frame capture hardware to frame capture software.I'm just not sold on it's use as an unbiased baseline to test from.
So benchmarks show crossfire working but you show absolutely no change from single GPU to dual. That just does not make sense. ANd what exactly is OBSERVED FPS anyways?Edited by Redwoodz - 9/19/13 at 11:48am