The NUCAR Army Publications General Information Conferences Research Projects Home

LEVO Runtime Predication Research Students: David Morano Recent Participants: Marcos de Alba Alireza Khalafi Faculty: David Kaeli Augustus Uht, URI Project Summary: We are proposing a machine model that realizes much of the potential ILP present in codes. Our approach, which we call Levo, speculates across multiple data and control dependencies, executing many instructions, and then re-executing some of them when mis-speculation occurs. Levo utilizes a hybrid form of the Tomasulo algorithm which we call resource flow. Levo does not rely on any added compiler or software support to obtain ILP. Related information can be found at: The Levo home page at URI. Also see the NUCAR publications page.