Lambda the Ultimate

Today I was thinking about the architectural differences between current/regular/normal/whatever systems, and the VLIW designs that Intel has come up with.
If you're not familiar with these systems (Itanium, etc), here are some good links:
IA-64 @ Wikipedia
IA-64 Preview @ ArsTechnica

It occurred to me that I had no idea how dynamic and interpreted languages like Python, Lisp, and the others, would perform on such systems.
Does assigning the bulk of the optimization work to the compiler help or hurt the performance of applications written in these languages?

I found an interesting thread on this topic here:
lists.tuned.org archives
..but it's 6 years old. Surely there's more information about these things now that they are widely available.

Is this still largely an unknown? I'm curious.