Application Topology Definition and Tasks Mapping for Efficient Use of Heterogeneous Resources

Abstract

Nowadays, high-performance computing (HPC) not only faces challenges to reach computing performance, it also has to take in consideration the energy consumption. In this context, heterogeneous architectures are expected to tackle this challenge by proposing a mix of HPC and low-power nodes. There is a significant research effort to define methods for exploiting such computing platforms and find a trade-off between computing performance and energy consumption. To this purpose, the topology of the application and the mapping of tasks onto physical resources are of major importance. In this paper we propose an iterative approach based on the exploration of logical topologies and mappings. These solutions are executed onto the heterogeneous platform and evaluated. Based on these results a Pareto front is built, allowing users to select the most relevant configurations of the application according to the current goals and constraints. Experiments have been conducted on a heterogeneous micro-server using a video processing application running on top of a software-distributed shared memory and deployed over a mix of Intel i7 and Arm Cortex A15 processors. Results show that some counterintuitive solutions found by the exploration approach perform better than classical configurations.

Adaptive message passing polling for energy efficiency: Application to software‐distributed shared memory over heterogeneous computing resources

“…In this work, we consider an arbitrary configuration for each application: the optimization of the deployment is addressed in another contribution. 16 ⊳ Experiment F: Reducing the S-DSM CPU load with real applications.…”

“…These methods include heuristics based on genetic algorithms, 21 design space exploration, 33 static scheduling optimization, 34 and local search. 16 The latter work is applied to distributed applications relying on the OpenMPI runtime and deployed onto heterogeneous machines. In all these systems, the application is monitored and a configuration is put into a Pareto front according to the energy consumption that has been measured.…”

“…The configuration of the deployment consists in several steps including sizing the application (how many instances of each task), the logical topology (how instances are connected), the mapping of the instances onto the physical resources. In this work, we consider an arbitrary configuration for each application: the optimization of the deployment is addressed in another contribution 16 …”

“…Some research efforts are made to build a trade‐off between computing performance and energy consumption when configuring distributed applications. These methods include heuristics based on genetic algorithms, 21 design space exploration, 33 static scheduling optimization, 34 and local search 16 . The latter work is applied to distributed applications relying on the OpenMPI runtime and deployed onto heterogeneous machines.…”

Software-Distributed Shared Memory for Heterogeneous Machines: Design and Use Considerations

“…These deployment steps are complex even when coping with rather small applications and small heterogeneous clusters. In this paper [20], a compilation toolchain is proposed to explore and optimize the deployment of the S-DSM onto distributed heterogeneous resources, using local search algorithm. This results in a Pareto front from which it is possible to pick a solution that fits to specific computing performance and energy consumption constraints.…”