Automatic Parallelizing Compilers have existed for several decades. The concept is simple: write a computer program to run on a single processor and let the compiler figure out how to compile it in a way that is suitable for execution on multiple processors. The problem is building such a compiler is very difficult and the research community has little to show for its effort after all this time. Although there have been plenty of success stories for specific cases, parallelizing compilers are rarely used and are unlikely to find wide acceptance anytime soon. So is it even worth trying to build parallelizing compilers?
In this talk, Dr. Ferner will discuss some recent research involving a parallelizing compiler he has built and the difficulties of automatic parallelization. He will discuss the growing need for automatic parallizing compilers. Dr. Ferner will also discuss what the recent trends in computing, such as Grid Computing, Cloud Computing, and Multi-Core processors mean for the field of parallelizing compilers.
