Idle waves in high-performance computing

Venčels, Juris; Akhmetova, Dana; Laure, Erwin; Henri, Pierre

doi:10.1103/physreve.91.013306

Cited by 19 publications

(17 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Data communication modeling in high-performance computing is a subject of constant development 50,51 . In particular for message-passing parallel programming models where blocking, as well as, nonblocking communications while synchronizing nearest-neighbor processes affect the performance of the simulations 52 .…”

Section: A Modeling and Scaling Laws For Hespaddamentioning

confidence: 99%

Scalable and fast heterogeneous molecular simulation with predictive parallelization schemes

et al. 2017

View full text Add to dashboard Cite

Multiscale and inhomogeneous molecular systems are challenging topics in the field of molecular simulation. In particular, modeling biological systems in the context of multiscale simulations and exploring material properties are driving a permanent development of new simulation methods and optimization algorithms. In computational terms, those methods require parallelization schemes that make a productive use of computational resources for each simulation and from its genesis. Here, we introduce the heterogeneous domain decomposition approach which is a combination of an heterogeneity sensitive spatial domain decomposition with an a priori rearrangement of subdomainwalls. Within this approach, the theoretical modeling and scaling-laws for the force computation time are proposed and studied as a function of the number of particles and the spatial resolution ratio. We also show the new approach capabilities, by comparing it to both static domain decomposition algorithms and dynamic load balancing schemes. Specifically, two representative molecular systems have been simulated and compared to the heterogeneous domain decomposition proposed in this work. These two systems comprise an adaptive resolution simulation of a biomolecule solvated in water and a phase separated binary Lennard-Jones fluid.

show abstract

Section: A Modeling and Scaling Laws For Hespaddamentioning

confidence: 99%

Scalable and fast heterogeneous molecular simulation with predictive parallelization schemes

et al. 2017

View full text Add to dashboard Cite

show abstract

“…Previous works on simulations of idle period propagation showed that idle period propagates among processes in certain point-to-point communication patterns such as neighbor communication [1] and collective operations [3]. LogGOPSim simulations [4] predicted that idle periods propagate among MPI processes as waves with a propagation speed depending on the average computational time and network parameters [1]. Montecarlo simulations showed that certain classes of collective operations have the property of hiding process imbalance and minimizing idle period propagation [3].…”

Section: Related Workmentioning

confidence: 99%

“…In a different work based on simulation results, [6] pointed out that both collective and point-to-point operations can impact the sensitivity of parallel applications to system noise and that the blocking communication mode has an amplification effect of system noise. Also based on simulation results, [1] presented an equation description of the propagation speed of idle waves in scientific applications on large number of processes.…”

Section: Related Workmentioning

confidence: 99%

“…Simulations of MP applications showed that idle periods are generated on a process when it needs to wait for a delayed process. This idle period can propagate among different processes as a wave [1]. An example of such mechanism is shown in Figure 1 that represents an application running on seven processes with data exchange only among neighbor processes.…”

Section: Introductionmentioning

confidence: 99%

“…In MP applications, the delay on one process can result in an idle period on the processes waiting for data from this delayed process. This implicit local synchronization of point-to-point communication allows for the propagation of the idle period among communicating processes [1], [3]. 2) Shared resources.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Idle Period Propagation in Message-Passing Applications

Laure

Kestor

Gioiosa

2016

2016 IEEE 18th International Conference on High Performance Computing and Communications; IEEE 14th International Conference On

Self Cite

View full text Add to dashboard Cite

Abstract-Idle periods on different processes of Message Passing applications are unavoidable. While the origin of idle periods on a single process is well understood as the effect of system and architectural random delays, yet it is unclear how these idle periods propagate from one process to another. It is important to understand idle period propagation in Message Passing applications as it allows application developers to design communication patterns avoiding idle period propagation and the consequent performance degradation in their applications. To understand idle period propagation, we introduce a methodology to trace idle periods when a process is waiting for data from a remote delayed process in MPI applications. We apply this technique in an MPI application that solves the heat equation to study idle period propagation on three different systems. We confirm that idle periods move between processes in the form of waves and that there are different stages in idle period propagation. Our methodology enables us to identify a selfsynchronization phenomenon that occurs on two systems where some processes run slower than the other processes.

show abstract

Key to High Performance Ion Hybrid Capacitor: Weakly Solvated Zinc Cations

Chen,

Sun,

Plietker

et al. 2023

Advanced Science

View full text Add to dashboard Cite

Zinc ion hybrid capacitors suffer from lack of reversibility and dendrite formation. An electrolyte, based on a solution of a zinc salt in acetonitrile and tetramethylene sulfone, allows smooth zinc deposition with high coulombic efficiency in a Zn||stainless steel cell (99.6% for 2880 cycles at 1.0 mA cm−2, 1.0 mAh cm−2). A Zn||Zn cell operates stably for at least 7940 h at 1.0 mA cm−2 with an area capacity of 10 mAh cm−2, or 648 h at 90% depth of discharge and 1 mA cm−2, 9.0 mAh cm−2. Molecular dynamics simulations reveal the reason for the excellent reversibility: The zinc cation is only weakly solvated than in pure tetramethylene sulfone with the closest atoms at 3.3 to 3.8 Å. With this electrolyte, a zinc||activated‐carbon hybrid capacitor exhibits an operating voltage of 2.0 to 2.5 V, an energy‐density of 135 Wh kg−1 and a power‐density of 613 W kg−1 at 0.5 A g−1. At the very high current‐density of 15 A g−1, 29.3 Wh kg−1 and 14 250 W kg−1 are achieved with 81.2% capacity retention over 9000 cycles.

show abstract

Idle waves in high-performance computing

Cited by 19 publications

References 3 publications

Scalable and fast heterogeneous molecular simulation with predictive parallelization schemes

Scalable and fast heterogeneous molecular simulation with predictive parallelization schemes

Idle Period Propagation in Message-Passing Applications

Key to High Performance Ion Hybrid Capacitor: Weakly Solvated Zinc Cations

Contact Info

Product

Resources

About