|The stupid question is the question not asked|
Personally, I think Will_the_Chill has done an amazing job in 17 days of getting/digesting/presenting a lot of interesting information about a "Perl 5 Optimizing Compiler". Whether the outcome is 5% or 300% or whatever, only time will tell.
Will_the_Chill introduced himself as a businessman, CTO of an Austin company, and has the business need to see this succeed. I personally think that the success of this project could move Perl back into the mainline of IT.
Before even seeing a single line of Perl code, I sat in meetings where different languages were discussed, and it was the bias against Perl that made me wait to start using it. On the negative side was always the lack of a 'compiler'. I have found lots of ancient posts on PM and elsewhere for the need for a compiler and all the 'technical' reasons why it won't work!
But maybe this time it will work for business reasons.
I think the concept of the LLVM backend to be exciting, and reading the pdf that BrowserUk referenced, it seems like a viable possibility. If I read it correctly, what I liked was that LLVM could operate in different modes, and when optimization wasn't possible, it defaulted back to the interpreter mode. That capability may help in the area of 'magic' vs 'not magic' that has repeatedly been brought up in this discussion.
Just my 2 cents!
"Well done is better than well said." - Benjamin Franklin