I was the one who performed that brief testing of the various algorithms, I can tell you that the results were note wrong, BUT from looking into this further, I can see why they appear that way. The testing I did was with running 2 threads with an intensity of 13 (increasing intensity had no affect), I did this, as with all current algorithms I had tested at that point, my cards performed better on 2 threads compared to 1, and 4800Khs was the max I got from them.
I have changed the parameters to 1 thread with intensity 19, and I do get ~5900Khs from them now.

I'll go back and update the results I got in the DMD forum when I have time, to reflect these findings.

"Algorithm: 4 rounds of hashing from fugue -> shavite -> hamsi -> panama (ASIC resistent!)"
A mish-mash of science algo appraoch. Thats a FAIL.
Energy efficiency only takes you so far esp if difficulty increases proportionally to energy efficiency (i.e. status quo)

The key feature of Groestlcoin is its good performance on a wide range of platforms and reduced ASIC advantages the algo contains. This  is in part due to the Grøstl algo's resistance to parallelization.

Mish-mash science PoW's wont work similarly well on different platforms.

Would have done better to clone Groestl :p

Simply put Groestl is the most democratic PoW for crypto mining