Originally Posted by specopsFI
Maybe because we have a review mentioned in the thread title that shows otherwise. And we haven't seen anything to show how that review is flawed. Yes, you can get way better fps numbers or benching results with an OC'd 7950, but if it has less stable frame times, then that there is a perfectly good reason to choose GTX 660Ti. I wouldn't, but if someone was to make that choice based on latency numbers, I'd understand it perfectly well. Everyone doesn't have same priorities as me.
So latency numbers or frame times not in line with Tech Report, if you please.
1. The problem is with measuring frametime at the GPU instead of the monitor where its perceived.
2. Also I have a simple question . Take 2 cards for instance. one having frametimes alternating between 10 ms and 40ms for every frame. the second card having 33.3 ms constant frametime per frame. The first card takes 50 ms for every 2 frames (20 such pairs of frames per sec) while the second card takes 66.6 ms for every 2 frames(15 such pairs of frames per sec). the first card gets 40 fps. the second card gets 30 fps. According to TR's method the first card would have 134ms spent in long frames per second ( 20 x (40 - 33.3) = 134 ms) ie frames which have frametime above the 33.3 ms threshold. The second card will have 0 ms above the 33.3ms threshold.
Now the issue is the first card runs at 40 fps . the second card runs at 30 fps. so which will the user perceive as smoother. Is it the first card or the second card. Can the user perceive the first card to be less smoother when in 50 ms the card has already rendered 2 frames while the second card has only rendered 1 frame and has not yet rendered the second frame ?Edited by raghu78 - 12/12/12 at 9:41am