Modeling energy consumption of parallel applications

Abstract

Abstract-The paper presents modeling and simulation of energy consumption of two types of parallel applications: geometric Single Program Multiple Data (SPMD) and divide-and-conquer (DAC). Simulation is performed in a new MERPSYS (Modeling Efficiency, Reliability and Power consumption of multilevel parallel HPC SYStems using CPUs and GPUs) environment. Model of an application uses the Java language with extensions representing message exchange between processes working in parallel. Simulation is performed by running threads representing distinct process codes of an application, with consideration of process counts. Instead of running time consuming calculations, their times are simulated using functions representing computational time dependent on input data sizes. The simulator considers performance and power consumption values for compute devices stored in its database. We performed verification of running the two applications on up to 512 and 1024 processes respectively on a large cluster from Academic Computer Center in Gdansk demonstrating a high degree of accuracy between simulated and measured results.

Energy-Aware High-Performance Computing: Survey of State-of-the-Art Tools, Techniques, and Environments

“…e tool provides means (Java scripts specified using the web simulator interface) for the flexible system and application definitions for simulating energy consumption and the execution time. e simulator was tested using typical SPMD (Simple Program Many Data) and DAC (Divide and Conquer) applications [82].…”

“…GSSim/DCworms Grid [80,81] A scheduler simulation concerning performance and energy consumption for complex grid architectures MERPSYS Grid/cluster/compute node [82,83] Used to simulate the energy consumption of a cluster compute nodes CloudSim Cloud [84] Used for simulation of VM provisioning in a cloud environment SimGrid Grid [85] Focused on its versatility and scalability GENSim Cloud [86] Used to simulate green energy prediction OMNet++, INET Cluster [58] Used for simulation of switching off the unused nodes in a cluster GDCSim Data center [87] Used for holistic evaluation of HPC and Internet data centers GreenCloud Data center [88] Used for evaluation of cloud data centers with carious infrastructure architectures TracSim Cluster [89] Used for maximizing the performance for a given power cap ASKALON Cloud [90] Used for cloud simulation with a given power cap Energy-aware HyperSim-G Grid [74] Used for assessment of energy-aware scheduling algorithms GPU design space exploration GP GPU [39] Dedicated for multiobjective GP GPU evaluation and selection Sniper + McPAT CPU [35] Used for multicore CPU energy-aware simulation with heuristics. e simulation was based on the Sniper [99] package, aiming to increase efficiency by optimizing the level of the simulation accuracy.…”

Analyzing energy/performance trade-offs with power capping for parallel applications on modern multi and many core processors

“…The taxonomy of methods considered in this work includes power-aware scheduling and resource management, parallelization oriented including balancing, communication focused, approximation methods with part of computations executed with lower energy usage and slight loss of accuracy. In papers [6] and [7] we modeled energy consumption of parallel applications focusing on various communication routines establishing both functions and coefficients valid for various cluster systems.…”

Modelling and simulation of GPU processing in the MERPSYS environment

“…MERPSYS was successfully used by its authors to model execution time, energy consumption and reliability of divide-and-conquer (DAC [18,13]) and geometric single program multiple data (SPMD [13]) applications [16], K-means algorithm [17] and volunteer computing systems [14]. For a more detailed description providing better insight into the whole MERPSYS environment see [15].…”