Speculative execution is a powerful technique for exploring the full potential of wideissue superscalar and very long instruction word (VLIW) processors. There are two classes of speculative execution, control speculation and data speculation. Control speculation refers to the execution of operations before it has been determined that they would be executed in the normal control ow of execution. The concept of control speculation has long been recognized for its ability to improve code performance [1]. Data speculation involves executing load instructions before possibly aliased operations such as stores. Although similar in nature to control speculation, the concepts concerning data speculation are just now maturing. In this thesis, a research infrastructure is presented for supporting data speculative optimizations and scheduling.