use unordered_map for better performance

[piotr-sikora-v]

I tried it on X5650 genproclimit=20 increase from 6100 hps to 6500hps

[biblepay]

Oh my, OK thanks a lot.

I will open the floodgates very soon. Working on a mining-for-cancer issue now, give me a little bit of time to look at this locally.

[piotr-sikora-v]

for x5650 xeon is about 5 to 10 % I just searched for better functions to scan, and find many performance sttistics that show it unorderer_map is faster than other hash tables. example: http://incise.org/hash-table-benchmarks.html

[piotr-sikora-v]

and this one: http://scottmeyers.blogspot.de/2015/09/should-you-be-using-something-instead.html

[biblepay]

Hmmmm... Not sure if we should induct this, does it appear to be harmless?