|parallel code optimization? email@example.com (1993-01-20)|
|parallel code optimization firstname.lastname@example.org (1993-01-21)|
|From:||email@example.com (John Kapenga)|
|Organization:||Western Michigan Univ. Comp. Sci. Dept.|
|Date:||Wed, 20 Jan 1993 07:21:05 GMT|
|Keywords:||parallel, optimize, question|
Does anyone have any references on parallelization of code optimization?
Or, a related question, suppose you had 10,000 times more processing
power. What would you do differently in the gcc optimizer?
One approach would be to use some exhaustive search, such as the GNU Super
Optimizer. However, a 10,000 fold speedup might not be enough for that on
any large scale. :-)
I have a student who is interested in moving some rtl optimizations from
the middle of gcc to a parallel machine and back.
How much room is left for additional optimization (without
inter-procedural analysis) on currrent compilers? From looking at SPARC
compilers (SUN and gcc) I might suppose 10-20% is still there without
getting real strange.
John A. Kapenga Department of Computer Science
Ma Bell: (616)387-5657 Western Michigan University
Internet: firstname.lastname@example.org Kalamazoo, MI 49008
Return to the
Search the comp.compilers archives again.