I'm still skeptic about 64bits computer architecture, memory transfers are already 64bits since the pentium 3, and there's really little use for numbers as large as 64 bit integers for common applications. Floating point computing is more intensive these days anyway.
I must admit that I've been lazy as of late to keep myself updated in the subject, I'm just talking from an engineer POV, the reasons for the jump to 64bits are beyond my current knowledge, they might be reasonable but they could be just a bunch of marketing bs.