Using an adaptive and time predictable runtime system for power-aware HPC-oriented applications

Portero, Antoni; Ševčı́k, Jiřı́; Golasowski, Martin; Vavrik, Radim; Libutti, Simone; Massari, Giuseppe; Catthoor, Francky; Fornaciari, William; Vondrák, Vít

doi:10.1109/igcc.2016.7892621

Cited by 3 publications

(4 citation statements)

References 2 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this section, we run the uncertainty module on 16 nodes of an HPC cluster. The module uses a hybrid OpenMP and MPI approach [16,21,22] to distribute the computations to multiple nodes. Given that the performance requirements of the application are time-variable (e.g., low when sunny, intermediate/critical when rainy), the HPC center may allocate to the application only some computing nodes and use the remaining ones to execute other applications.…”

Section: Methodsmentioning

confidence: 99%

Just-In-Time Execution Through On-Demand Resource Allocation in HPC Systems

Portero¹,

Podhoranyi²,

Hrbac³

et al. 2017

Proceedings of the 1st International Conference on Algorithms, Computing and Systems

Self Cite

View full text Add to dashboard Cite

This article is centred on a mathematical weather forecasting model that must run regularly (i.e. 24/7) on an HPC system. Depending on the environmental conditions, each execution of the model may have a different deadline and a different accuracy requirement. In order to minimize power consumption and heat, we minimize resource allocation as far as the deadlines allow, thus evenly spreading resource usage over time while nonetheless complying with the deadlines. Our work relies on a run-time resource manager that adapts resource allocation to the runtimevariable performance demand of applications. The resource assignment is temperature-aware: the application is dynamically migrated on the coolest cores, and this has a positive impact on the system reliability. CCS Concepts• Computing methodologies ➝ Simulation evaluation • Computing methodologies ➝ Simulation environments • Computing methodologies ➝ Concurrent algorithms.

show abstract

Section: Methodsmentioning

confidence: 99%

Just-In-Time Execution Through On-Demand Resource Allocation in HPC Systems

Portero¹,

Podhoranyi²,

Hrbac³

et al. 2017

Proceedings of the 1st International Conference on Algorithms, Computing and Systems

Self Cite

View full text Add to dashboard Cite

show abstract

“…In the case of the HARPA-OS, the runtime always provides the minimized resources to execute the models in the given time frame (before the deadline). The study [22] shows that the HARPA-OS mechanism has advantages regarding power consumption and a decrease in the temperature of the cores. Since total runtime is limited due to the QoS and the goal is to minimize power and heat.…”

Section: A Disaster and Flood Management Simulationmentioning

confidence: 96%

“…The application is used to simplify the process of disaster management and increase its operability and effectiveness. POLIMI-IT4I have developed a case study [22] to evaluate the resource allocation capabilities of the HARPA-OS on an HPC machine equipped with an x86-64 multi-core machine. The case study simulates the uncertainty (Uncer) of the rainfall-runoff (RR) model.…”

Section: A Disaster and Flood Management Simulationmentioning

confidence: 99%

See 1 more Smart Citation

HARPA: Tackling physically induced performance variability

Zompakis¹,

Noltsis²,

Ndreu

et al. 2017

Design, Automation &Amp; Test in Europe Conference &Amp; Exhibition (DATE), 2017

View full text Add to dashboard Cite

Continuously increasing application demands on both High Performance Computing (HPC) and Embedded Systems (ES) are driving the IC manufacturing industry on an everlasting scaling of devices in silicon. Nevertheless, integration and miniaturization of transistors comes with an important and nonnegligible trade-off: time-zero and time-dependent performance variability. Increasing guard-bands to battle variability is not scalable, since worst-case design margins are prohibitive for downscaled technology nodes. This paper discusses the FP7-612069-HARPA project of the European Commission which aims to enable next-generation embedded and high-performance heterogeneous many-cores to cost-effectively confront variations by providing Dependable-Performance: correct functionality and timing guarantees throughout the expected lifetime of a platform under thermal, power, and energy constraints. The HARPA novelty is in seeking synergies in techniques that have been considered virtually exclusively in the ES or HPC domains (worstcase guaranteed partly proactive techniques in embedded, and dynamic best-effort reactive techniques in high-performance).

show abstract