Taskgraph: A Low Contention OpenMP Tasking Framework

TL;DR

taskgraph is a framework that reduces OpenMP task management overheads using a dependency graph, enabling efficient fine-grained parallelism and outperforming standard OpenMP implementations in performance and scalability.

Contribution

introduces taskgraph, a novel dependency graph-based framework that significantly lowers OpenMP runtime overheads and enhances fine-grained task parallelism.

Findings

outperforms vanilla OpenMP in performance and scalability

reduces overheads related to task dependency resolution

enables effective exploitation of fine-grained parallelism

Abstract

OpenMP is the de-facto standard for shared memory systems in High-Performance Computing (HPC). It includes a task-based model that offers a high-level of abstraction to effectively exploit highly dynamic structured and unstructured parallelism in an easy and flexible way. Unfortunately, the run-time overheads introduced to manage tasks are (very) high in most common OpenMP frameworks (e.g., GCC, LLVM), which defeats the potential benefits of the tasking model, and makes it suitable for coarse-grained tasks only. This paper presents taskgraph, a framework that uses a task dependency graph (TDG) to represent a region of code implemented with OpenMP tasks in order to reduce the run-time overheads associated with the management of tasks, i.e., contention and parallel orchestration, including task creation and synchronization. The TDG avoids the overheads related to the resolution of task dependencies and greatly reduces those deriving from the accesses to shared resources. Moreover, the taskgraph framework introduces in OpenMP the record-and-replay execution model that accelerates the taskgraph region from its second execution. Overall, the multiple optimizations presented in this paper allow exploiting fine-grained OpenMP tasks to cope with the trend in current applications pointing to leverage massive on-node parallelism, fine-grained and dynamic scheduling paradigms. The framework is implemented on LLVM 15.0. Results show that the taskgraph implementation outperforms the vanilla OpenMP system in terms of performance and scalability, for all structured and unstructured parallelism, and considering coarse and fine grained tasks. Furthermore, the proposed framework considerably reduces the performance gap between the task and the thread models of OpenMP.