Who can prove the veracity of any vendor claim? I don't want to get into that.
I can tell you that the AES-NI (AES-GCM) changes will blow the doors off AES-CBC (what you're seeing now).
http://freebsdfoundation.blogspot.com/2014/08/freebsd-foundation-announces-ipsec.html
I can also tell you that these changes are being tested against a C2758 and another, different Rangeley board.
That said, they should work on the i7 just as well.
Also, while it isn't running today, there is a "QuickAssist" part on the C2758 that will eventually run (I've just re-engaged Intel
about it.) When that code is finished (and in pfSense), it will blow the doors off the i7 as far as crypto is concerned.
https://www.youtube.com/watch?v=M49TKu2cx-Q
http://lkml.iu.edu/hypermail/linux/kernel/1406.0/01810.html
https://01.org/packet-processing/intel®-quickassist-technology-drivers-and-patches
(I might get the regex stuff going as well, which could help (a lot) with Snort.
http://marc.info/?l=snort-devel&m=128396544311154&w=2
Quad core @ 3.4GHz, or 8 core @ 2.4GHz? Hmm.
The price for the i7 (80 SSD, 8GB ram, single PSU) on that site is $1,830.35. The C2758 is $1500.
Both are supported by the vendor, though the vendor for the C2758 is pfSense.