The Parallel Semantics Program Dependence Graph

TL;DR

The paper introduces the Parallel Semantics Program Dependence Graph ($PS-PDG$), a generalization of the Program Dependence Graph designed to capture the exact parallel constraints of OpenMP and Cilk programs. By adding hierarchical nodes, node traits, contexts, data-selectors, and parallel semantic variables, PS-PDG enables compilers to explore a wider space of semantically-equivalent parallel execution plans while preserving semantics. The authors prove the necessity and sufficiency of PS-PDG for OpenMP semantics, implement it in a NOELLE-based compiler, and show that it expands optimization options and improves potential parallelism compared to PDG-based approaches. The results indicate that PS-PDG can unlock substantial speedups on many-core architectures by decoupling parallel semantics from the programmer-encoded parallelization plan. This abstraction has practical impact by enabling more aggressive and semantically correct parallel optimizations in modern compilers.

Abstract

A compiler's intermediate representation (IR) defines a program's execution plan by encoding its instructions and their relative order. Compiler optimizations aim to replace a given execution plan with a semantically-equivalent one that increases the program's performance for the target architecture. Alternative representations of an IR, like the Program Dependence Graph (PDG), aid this process by capturing the minimum set of constraints that semantically-equivalent execution plans must satisfy. Parallel programming like OpenMP extends a sequential execution plan by adding the possibility of running instructions in parallel, creating a parallel execution plan. Recently introduced parallel IRs, like TAPIR, explicitly encode a parallel execution plan. These new IRs finally make it possible for compilers to change the parallel execution plan expressed by programmers to better fit the target parallel architecture. Unfortunately, parallel IRs do not help compilers in identifying the set of parallel execution plans that preserve the original semantics. In other words, we are still lacking an alternative representation of parallel IRs to capture the minimum set of constraints that parallel execution plans must satisfy to be semantically-equivalent. Unfortunately, the PDG is not an ideal candidate for this task as it was designed for sequential code. We propose the Parallel Semantics Program Dependence Graph (PS-PDG) to precisely capture the salient program constraints that all semantically-equivalent parallel execution plans must satisfy. This paper defines the PS-PDG, justifies the necessity of each extension to the PDG, and demonstrates the increased optimization power of the PS-PDG over an existing PDG-based automatic-parallelizing compiler. Compilers can now rely on the PS-PDG to select different parallel execution plans while maintaining the same original semantics.

Figures (14)

• Figure 1: Compiler pipeline utilizing IR and PDG for code optimization.
• Figure 2: Compiler pipeline utilizing parallel IR and PS-PDG for parallel code optimization.
• Figure 3: The key computational kernel from the IS benchmark with the original and a more performant compiler-selected parallel execution.
• Figure 4: Comparison of the existing pipeline with our proposed PS-PDG pipeline. The parallelization plan in the source code is abstracted away.
• Figure 5: Capturing properties of a region into hierarchical nodes with traits