YEs, basically the samples are a bit rubbish, high power, either the early samples are just test silicon of the cpu alone or the IGP is there but not working.
The later part describes what Ivybridge is intended to be as a final released product.
PCI-e 3 will be most likely, pointless for all but a few niche markets, insane bandwidth £15k a drive pci-e ssd drives with silly silly silly speed. Also current gpu's and current bandwidth it would just make for instance quadfire/quad sli a bit more manageable as 4 8x slots would offer the same bandwidth as 4 16x slots currently, the problem being mobo's with those kinds of setups need extra pci-e chips for more lanes and speed, so you could get away without them and on cheaper boards.
THe reality is though, almost no one quadfires/quad sli's, single/dual card setups will basically not benefit from pci-e 3.
Interesting chips, not expecting a massive amount architecturally and 30% faster IGP won't interest that many people. Hessler, kessler, whatever the next 22nm parts from Intel are, are looking far more interesting. Should be finally bringing 6/8 cores to the mainstream price bracket. So over Sandybridge Ivy is supposed to be 20% faster, and a lot of that seems to be coming from clockspeed and gpu vs overall speed increase, while the next thing will be going from 4 to probably 8 cores..... Ivy's not interesting, the next thing along is VERY interesting.