Is the main point here to contribute towards the aim of a CPU-only algorithm? I think this is pretty cool and I also want to understand the architecture of 'why' this would work to be better for a CPU.
Yes, we want to stay CPU-ONLY.
But I never claimed that any of the updates is "better" for a CPU.
The main goal behind BurgerHash is to keep on adding to the complexity of the hash algo so that any GPU-miners that are created will be obsolete after the next update.
The basic premise is that coding for GPUs is magnitudes for complex than coding for CPUs.Is there something that explains the theory behind why the Burgerhash algorithm is significantly more efficient on a CPU than any other computing platform?
Again, I never claimed that, I never used these words. Please only read what I write, not what others say about BurgerHash. (except of course when they have valid criticisms, which are always welcome)
Also, I do not think that such an algorithm even exists (which is provably more efficient on a CPU than on one of the more advanced GPUs. I know that some claims have been made, but I doubt those functions are valid for POW-hashing)
With BurgerHash I try to add various mathematical angles to the hash algo which will force GPU-miners to do more and more extra work to keep up.
So let me give you one example of my thought-process behind the StickyPuffyCheese-Upgrade:
I know for a fact that with changing size of the input-data you need to change a whole lot more in CUDA then you need when just using the KECCAK function of a library like say sph on CPU.
Padding sizes and what not needs to be specifically adjusted when using CUDA which is not necessary for mere C++ on CPU.
A well-versed GPU-dev might do exactly that, and he might be able to mine ZETTEL on GPU for a few weeks, until I throw some more extra shit his way.
That's the basic idea.
BTW: as POW-hash-research continues I might come up with advanced mechanisms that also decrease the amount of hash-benefit a GPU-miner might get compared to CPU. But this remains to be seen.