Bin packing looks better because I don't think SSA has been implemented yet. Linear Search looks perfect for SSA code, but Bin Packing would be better for Parrot. (Bin Packing views allocation as a bin packing problem and solves it heuristically.)
On Jun 29, 2006, at 11:20 PM, Vishal Soni wrote: Hi Everyone,
Currently IMCC uses a Graph Coloring based Register allocation algorithm. The implementation is a trimmed down version of Brigg's Allocator.
I came across this research paper that talks about the new register allocation algorithm "Linear Scan Allocation"for dynamically compiled languages. Parrot perfectly fits the mold of dynamically compiled language.
The Linear Scan Allocator is faster at register allocation process and seems to have the same execution time for the code. For more information please refer to the research paper from IBM on Liner Scan Allocation
Let me know what your thoughts are and would it be worth implementing this algorithm to see how it performs compared to graph coloring algorithm.
Please share your thoughts accordingly
-- Thanks, Vishal
Sincerely, Watson Ladd --- "Those who would give up Essential Liberty to purchase a little Temporary Safety deserve neither Liberty nor Safety." -- Benjamin Franklin
|
PGP.sig
Description: This is a digitally signed message part