we should put more analysis over the BWT algorithm

[timespaceuniverse]

We should put more detailed analysis of what is indeed the bottle neck of the BWT transform ?
As BWT is supposed to be anti-ASIC/GPU . There are several factors under the hood .

Cache bottle neck (l1 l2 l3)
Cache size limit or read/write bandwidth limit ?
DDR memory bottle neck
As ETHASH is indeed using the DDR memory bandwidth limit (that is why ETH is GPU mining friendly)
We have tested the BWT algorithm over 97 cores CPU . And we could clearly see that there is no hash-rate increment
at some point with the increment of cpu cores.

We should do more concrete analysis why this is the case and why it is anti-gpu mining.

neglect the blue line