only get .6khash with 16x1 which is same as 14x and 13x
Try 18x1 for a couple of seconds while watching how much VRAM is being used in GPU-Z, if less then ~2600, go higher, 20x1, 22x1, or try different combinations, like K4x4, K4x5, K5x5, and if it's not using enough VRAM, just increase a number.
@cbuchner1: It seems that we're way behind even at N factor 10 (eg. Applecoin) due to Keccak. In my case there's virtually no difference between N factor 4 and 10, getting heavily CPU bottlenecked in both cases with the same performance.