This is just my assumption, but it seems the x13mod kernel is optimized for GCN core cards, ie. 7xxx/R7/R9 series. The 5xxx/6xxx run VLIW core technologies, totally different than GCN.
That's irrelevant. Plenty of algos are mined with cards that were never 'optimized' for a particular algo. But thx for the bump. Maybe someone helpful will post now
LOL...It is relevant, because the GCN core is more like a cpu structure and VLIW is more just a graphic accelerator than compute type gpu. So, that why is easier to optimize GCN card, than VLIW. Even Nvidia Maxwell core is kind of copying AMD's GCN core. More vectorization coding is needed to assist the VLIW cards.