اشترك بالحزمة الذهبية واحصل على وصول غير محدود شمرا أكاديميا
تسجيل مستخدم جديدPerformance and Energy Trade-Offs for Parallel Applications on Heterogeneous Multi-Processing Systems
67
0
0.0
(
0
)
اسأل ChatGPT حول البحث
ﻻ يوجد ملخص باللغة العربية
This work proposes a methodology to find performance and energy trade-offs for parallel applications running on Heterogeneous Multi-Processing systems with a single instruction-set architecture. These offer flexibility in the form of different core types and voltage and frequency pairings, defining a vast design space to explore. Therefore, for a given application, choosing a configuration that optimizes the performance and energy consumption is not straightforward. Our method proposes novel analytical models for performance and power consumption whose parameters can be fitted using only a few strategically sampled offline measurements. These models are then used to estimate an applications performance and energy consumption for the whole configuration space. In turn, these offline predictions define the choice of estimated Pareto-optimal configurations of the model, which are used to inform the selection of the configuration that the application should be executed on. The methodology was validated on an ODROID-XU3 board for eight programs from the PARSEC Benchmark, Phoronix Test Suite and Rodinia applications. The generated Pareto-optimal configuration space represented a 99% reduction of the universe of all available configurations. Energy savings of up to 59.77%, 61.38% and 17.7% were observed when compared to the performance, ondemand and powersave Linux governors, respectively, with higher or similar performance.
قيم البحث
اقرأ أيضاً
We consider energy minimization for data-intensive applications run on large number of servers, for given performance guarantees. We consider a system, where each incoming application is sent to a set of servers, and is considered to be completed if
a subset of them finish serving it. We consider a simple case when each server core has two speed levels, where the higher speed can be achieved by higher power for each core independently. The core selects one of the two speeds probabilistically for each incoming application request. We model arrival of application requests by a Poisson process, and random service time at the server with independent exponential random variables. Our model and analysis generalizes to todays state-of-the-art in CPU energy management where each core can independently select a speed level from a set of supported speeds and corresponding voltages. The performance metrics under consideration are the mean number of applications in the system and the average energy expenditure. We first provide a tight approximation to study this previously intractable problem and derive closed form approximate expressions for the performance metrics when service times are exponentially distributed. Next, we study the trade-off between the approximate mean number of applications and energy expenditure in terms of the switching probability.
Repurposing automotive batteries to second-use battery energy storage systems (2-BESS) may have environmental and economic benefits. The challenge with second-use batteries is the uncertainty and diversity of the expected packs in terms of their chem
istry, capacity and remaining useful life. This paper introduces a new strategy to optimize 2-BESS performance despite the diversity or heterogeneity of individual batteries while reducing the cost of power conversion. In this paper, the statistical distribution of the power heterogeneity in the supply of batteries is considered when optimizing the choice of power converters and designing the power flow within the battery energy storage system (BESS) to maximize battery utilization. By leveraging a new lite-sparse hierarchical partial power processing (LS-HiPPP) approach, we show a hierarchy in partial power processing (PPP) partitions power converters to a) significantly reduce converter ratings, b) process less power to achieve high system efficiency with lower cost (lower efficiency) converters, and c) take advantage of economies of scale by requiring only a minimal number of sets of identical converters. The results demonstrate that LS-HiPPP architectures offer the best tradeoff between battery utilization and converter cost and had higher system efficiency than conventional partial power processing (C-PPP) in all cases.
In this paper, we consider scalable output and regulated output synchronization problems for heterogeneous networks of right-invertible linear agents based on localized information exchange where in the case of regulated output synchronization, the r
eference trajectory is generated by a so-called exosystem. We assume that all the agents are introspective, meaning that they have access to their own local measurements. We propose a scale-free linear protocol for each agent to achieve output and regulated output synchronizations. These protocols are designed solely based on agent models and they need no information about communication graph and the number of agents or other agent models information.
This paper studies synchronization of homogeneous and heterogeneous discrete-time multi-agent systems. A class of linear dynamic protocol design methodology is developed based on localized information exchange with neighbors which does not need any k
nowledge of the directed network topology and the spectrum of the associated Laplacian matrix. The main contribution of this paper is that the proposed protocols are scale-free and achieve synchronization for arbitrary number of agents.
Adaptive synchronization protocols for heterogeneous multi-agent network are investigated. The interaction between each of the agents is carried out through a directed graph. We highlight the lack of communication between agents and the presence of u
ncertainties in each system among the conventional problems that can arise in cooperative networks. Two methodologies are presented to deal with the uncertainties: A strategy based on robust optimal control and a strategy based on neural networks. Likewise, an input estimation methodology is designed to face the disconnection that any agent may present on the network. These control laws can guarantee synchronization between agents even when there are disturbances or no communication from any agent. Stability and boundary analyzes are performed. Cooperative cruise control simulation results are shown to validate the performance of the proposed control methods.
سجل دخول لتتمكن من نشر تعليقات
التعليقات
جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
