|"Probabilistic" optimizations firstname.lastname@example.org (1990-02-27)|
|From:||email@example.com (John Gateley)|
|Date:||27 Feb 90 16:11:02 GMT|
|Organization:||TI Computer Science Center, Dallas|
I am looking for references on compiler optimizations based on
the frequency of IF/CASE branches. Please email and I will post a summary.
[The original Fortran compiler had a FREQUENCY statement that let you
tell it the relative likelihood that each of the three branches of an IF
would be taken or the typical number of times that a DO loop would run. It
turned out to be so useless that it disappeared by the time they came up
with Fortran II. I haven't looked for more recent references, -John]
Return to the
Search the comp.compilers archives again.