Skip to Main Content
Most of the actual approaches to address the problem of serial code executing on parallel multi-core processors try to exploit the parallelism of the code during compile time. Such a solution is very efficient but places the burden on the programmer to know the code will be running on a parallel architecture. In addition, the programmer needs to know the details of the architecture in order to generate efficient parallel code.In this work we present an alternative approach, the layer disruption model, that improves the performance of serial code executing on parallel architectures. We present a small test case that shows the benefits of the model.
Date of Conference: 1-4 Dec. 2008