On Sun, Sep 29, 2013 at 2:56 PM, GHC <ghc-devs@haskell.org> wrote:

According to the results, sadly the 5% performance gain in compile time
was complete bogus. Sorry. =(

Does anyone have any insight into why?

How can a future benchmark avoid such bogusity? The only -- I might add, implicit -- rec is to run things twice.

-- Kim-Ee