Do you mean XMR stak is 20% faster than Claymore?
Yes. Read my previous comments and links.
No, XMR-STAK do NOT outperform CCNM by 20%. At least not with Terascale GPUs. It is not easy even to get comparable performance. The problem is that CCNM wil not use Terascale any more after v7 fork.
Again, read my previous comments and links. You can achieve parity with XMR stak and even outperform CCNM precisely on TeraScale. It depends if you find the right intensity and worksize. Also the main problem with STAK and TeraScale was that it only detected 384mb of memory which limited every single card, now that has been addressed in this issue through my help and the help of other contributors
https://github.com/fireice-uk/xmr-stak/issues/472 I got 6990s doing ~395-400h/s per core, ~800h/s per gpu total on stak, vs 374 per core or ~ 750h/s on CCNM. Core clock and memory clocks identical in both cases.
It's no problem that CCNM will not support terascale anymore, problem is people are too lazy to work out a solution for themselves.
I use to run CCNM 9.7 400H/s+ and never ever seen more then 395 from XMR-STAK. What 20%+ are you speaking about??? Even you own data show only 6% that is usually within measuring error.