MINES ParisTech

The goal of this thesis is to design algorithms that run with better complexity when compiling or parallelizing loop programs. The framework within which our algorithms operate is the polyhedral model of compilation which has been successful in the design and implementation of complex loop nest optimizers and parallelizing compilers. The algorithmic complexity and scalability limitations of the above framework remain one important weakness. We address it by introducing sub-polyhedral compilation by using (Unit-)Two-Variable-Per-Inequality or (U)TVPI Polyhedra, namely polyhedra with restricted constraints of the type axi + bxj ≤ c (±xi ± xj ≤ c). A major focus of our sub-polyhedral compilation is the introduction of sub-polyhedral scheduling, where we propose a technique for scheduling using (U)TVPI polyhedra. As part of this, we introduce algorithms that can be used to construct under-aproximations of the systems of constraints resulting from affine scheduling problems. This technique relies on simple polynomial time algorithms to under-approximate a general polyhedron into (U)TVPI polyhedra. The above under-approximation algorithms are generic enough that they can be used for many kinds of loop parallelization scheduling problems, reducing each of their complexities to asymptotically polynomial time. We also introduce sub-polyhedral code-generation where we propose algorithms to use the improved complexities of (U)TVPI sub-polyhedra in polyhedral code generation. In this problem, we show that the exponentialities associated with the widely used polyhedral code generators could be reduced to polynomial time using the improved complexities of (U)TVPI sub-polyhedra. The above presented sub-polyhedral scheduling techniques are evaluated in an experimental framework. For this, we modify the state-of-the-art PLuTo compiler which can parallelize for multi-core architectures using permutation and tiling transformations. We show that using our scheduling technique, the above under-approximations yield polyhedra that are non-empty for 10 out of 16 benchmarks from the Polybench (2.0) kernels. Solving the under-approximated system leads to asymptotic gains in complexity, and shows practically significant improvements when compared to a traditional LP solver. We also verify that code generated by our sub-polyhedral parallelization prototype matches the performance of PLuTo-optimized code when the under-approximation preserves feasibility. ix te l-0 08 18 76 4, v er si on 1 29 A pr 2 01 3 www.dedoimedo.com all rights reserved x te l-0 08 18 76 4, v er si on 1 29 A pr 2 01 3