Appendix

For each program, the various combination of the optimization passes are listed below along with the results. The first number represents the number of SimpleSUIF instructions generated in the output code, and the second number is the actual number of instructions executed.

• O : Original
• C : Global constant propagation
• V : Local optimization (local common sub-expression elimination and local constant folding)
• L : Loop-invariant code motion
• D : Dead code elimination

swm256

ora

tomcatv128

linpackd200

mgrid16

compress