Some of the material in is restricted to members of the community. By logging in, you may be able to gain additional access to certain collections or items. If you have questions about access or logging in, please use the form on the Contact Page.
Loop unrolling is a compiler optimization that can improve the performance of applications without explicit intervention from programmers. In this study, we analyze how traditional loop unrolling techniques attempt to improve performance, and we propose new unrolling strategies that can unroll a greater number of loops than what is usually possible. We perform loop unrolling using our assembly optimizer, which analyzes and optimizes code at the assembly level instead of at the compiler level. This gives us a different perspective of the code than what a typical compiler would have, as we do not concern ourselves with the source code or an intermediate language. We are free to make modifications at the instruction level, which allows us to optimize large sections of code that might otherwise be impossible to work on. The optimizer collects statistics on the loops it unrolls, some of which are shown in our Results section. We ran our optimizer's unrolling algorithms on the SPEC 06 benchmark suite, and present both the statistics it generated as well as performance results from three different simulators. While performance gains are highly dependent on the application being optimized, we have shown it is possible to achieve considerable improvements by the use of loop unrolling.
