ScalOMP: Analyzing the Scalability of OpenMP Applications

Abstract

Achieving good scalability from parallel codes is becoming increasingly difficult due to the hardware becoming more and more complex. Performance tools help developers but their use is sometimes complicated and very iterative. In this paper we propose a simple methodology for assessing the scalability and for detecting performance problems in an OpenMP application. This methodology is implemented in a performance analysis tool named ScalOMP that relies on the capabilities of OMPT for analyzing OpenMP applications. ScalOMP reports the code regions with scalability issues and suggests optimization strategies for those issues. The evaluation shows that ScalOMP incurs low overhead and that its suggestions lead to significant performance improvement of several OpenMP applications.

One core dedicated to MPI nonblocking communication progression? A model to assess whether it is worth it

“…Though this simple approximation is sufficient for our model, as we will see in Section V, state-of-the-art on the scalability of OpenMP applications [23], [24] can be used to further improve the precision of our model.…”

sOMP: Simulating OpenMP Task-Based Applications with NUMA Effects

“…Task-based applications are indeed increasingly common, but overheads in runtime systems implementations may limit the applicability of the task model [12]. It is fundamentally important to be able to exhibit the precise performance of an OpenMP task-based application without artifacts from the runtime implementation: in addition to performance profiling [7,11], simulation is a way to achieve this goal. In previous works [25,24], we explored the simulation of task-based scheduling on heterogeneous architectures, which is now used as a reliable tool for scheduling experiments [1].…”