The polyhedral model is a well-known compiler optimization framework for the analysis and transformation of affine loop nests. We present a new method to solve a difficult geometric operation that is raised by this model: the integer affine transformation of parametric Z-polytopes. The result of such a transformation is given by a worst-case exponential union of Z-polytopes. We also propose a polynomial algorithm (for fixed dimension), to count points in arbitrary unions of a fixed number of parametric Z-polytopes. We implemented these algorithms and compared them to other existing algorithms, for a set of applications to loop nest analysis and optimization. © 2012 ACM.
CITATION STYLE
Seghir, R., Loechner, V., & Meister, B. (2012). Integer affine transformations of parametric Z-polytopes and applications to loop nest optimization. Transactions on Architecture and Code Optimization, 9(2). https://doi.org/10.1145/2207222.2207224
Mendeley helps you to discover research relevant for your work.