Release Dates - the only guarantee is the day you can buy one. I recall the LGA 2011 release date debate - I happened to be correct and some very good members insisted 2012 - neither of us knew for sure and relied on our sources
If you read or reread my first post and other posts I've had on this subject -- the BIG gains & benefits to PCIe 3.0 are not the GPUs themselves but the Chipsets and peripherals performance. The 'shared' bandwidth 'today' is a big problem with 'Swiss-Army' MOBO for USB, SATA, add-on Chips, etc and to a much lesser degree the 'GPU' itself. {if eSATA this then SATA that is disabled or Marvell SATA with x1 PCIe on SATA3 delivering 360~380MB/s vs ~600MB/s as 'expected'.
IMO - I expect very little to no GPU improvement by PCIe 2.x vs PCIe 3.0 alone. Not to be confused with a Faster GPU being Faster regardless to PCIe type. I mean if you example had: GTX 590 (PCIe 2.1) on a PCIe 2.0 x16 compared to GTX 590 (PCIe 3.0) on a PCIe 3.0 x16 = same performance and any differences would be <1%~0% and otherwise within a 'Margin of Error'. The reason is simple, the GTX 590 barely saturates the x8 and cannot saturate x16 PCIe 2.0.
What we needed 'yesterday' is 'SATA4' or whatever name sticks; the SSD's are and have been fully saturating the SATA3. Yes, I've seen the 'SATA Express' from 'SATA I-O'.