Little reflexion about pricing and FE popular discontent 1h before release.
Back at 2011 I bought one of the best AIB version of GTX560ti for 184.
It was based on a full GF114 chip, 384 cores, 1GB of RAM on a 256 bit bus, 360mm2 die size.
Absolute high end part from Nvidia was GTX580, based on full GF110 chip, 512 cores, 1,5GB of RAM on a 384 bit bus, 520mm2 die size.
So I got 75% of top end card cores, 66,67% of RAM, 66,67% of bus width and 65% of total transistor count. At least that chip got +6,48% higher default clocks to alleviate the difference. More when factoring OC on both cards.
Now, with 1060, I would have to pay around 350 for 50% cuda cores coming from 1080, with lower clocks on core and much lower clocks on memory on a 200mm2 die. But thing becomes absolutely terrible when we use the real, still unreleased, top end chip from Nvidia, reserved for Titan/ti. It should be based on GP102, with 384 bit bus if they can't manage to use HBM yet, 1060 would sport 33,33% of cores, 50% of memory at best considering a 12GB ti, 50% of bus width on top much lower memory clock and 41,84% of die size.
Scam falls short. But it is what it is with the beggar competence we have to suffer.