Search results for: schedulers

Authors: Subodh Kumar, Amit Varde

Abstract:

Typical on-premises high-performance computing (HPC) environments consist of a fixed number and a fixed set of computing hardware. During the design of the HPC environment, the hardware components, including but not limited to CPU, Memory, GPU, and networking, are carefully chosen from select vendors for optimal performance. High capital cost for building the environment is a prime factor influencing the design environment. A class of software called “Job Schedulers” are critical to maximizing these resources and running multiple workloads to extract the maximum value for the high capital cost. In principle, schedulers work by preventing workloads and users from monopolizing the finite hardware resources by queuing jobs in a workload. A cloud-based HPC environment does not have the limitations of fixed (type of and quantity of) hardware resources. In theory, users and workloads could spin up any number and type of hardware resource. This paper discusses the limitations of using traditional scheduling algorithms for cloud-based HPC workloads. It proposes a new set of features, called “HPC optimizers,” for maximizing the benefits of the elasticity and scalability of the cloud with the goal of cost-performance optimization of the workload.

Keywords: high performance computing, HPC, cloud computing, optimization, schedulers

Procedia PDF Downloads 92

Authors: Sina Saadati

Abstract:

Multiprocessors Systems-On-Chips (MPSOCs) are used widely on modern computers to execute sophisticated software and applications. These systems include different processors for distinct aims. Most of the proposed task schedulers attempt to improve energy consumption. In some schedulers, the processor's temperature is considered to increase the system's reliability and performance. In this research, we have proposed a new method for thermal-aware task scheduling which is based on an artificial neural network (ANN). This method enables us to consider a variety of factors in the scheduling process. Some factors like ambient temperature, season (which is important for some embedded systems), speed of the processor, computing type of tasks and have a complex relationship with the final temperature of the system. This Issue can be solved using a machine learning algorithm. Another point is that our solution makes the system intelligent So that It can be adaptive. We have also shown that the computational complexity of the proposed method is cheap. As a consequence, It is also suitable for battery-powered systems.

Keywords: task scheduling, MOSOC, artificial neural network, machine learning, architecture of computers, artificial intelligence

Procedia PDF Downloads 101

Authors: Rudolph Carl Barrientos, Juwaln Diego Descallar, Rainer James Palmiano

Abstract:

Household Appliance Scheduling Systems (HASS) coupled with a Time-of-Use (TOU) pricing scheme, a form of Demand Side Management (DSM), is not widely utilized in the Philippines’ residential electricity sector. This paper’s goal is to encourage distribution utilities (DUs) to adopt HASS and TOU by analyzing the effect of household schedulers on the electricity price and load profile in a residential environment. To establish this, a city based on an implemented survey is generated using Monte Carlo Analysis (MCA). Then, a Binary Particle Swarm Optimization (BPSO) algorithm-based HASS is developed considering user satisfaction, electricity budget, appliance prioritization, energy storage systems, solar power, and electric vehicles. The simulations were assessed under varying levels of user compliance. Results showed that the average electricity cost, peak demand, and peak-to-average ratio (PAR) of the city load profile were all reduced. Therefore, the deployment of the HASS and TOU pricing scheme is beneficial for both stakeholders.

Keywords: appliance scheduling, DSM, TOU, BPSO, city-wide simulation, electric vehicle, appliance prioritization, energy storage system, solar power

Procedia PDF Downloads 97

Authors: Hassen Hamouda, Mohamed Ouwais Kabaou, Med Salim Bouhlel

Abstract:

The growth of wireless technology made opportunistic scheduling a widespread theme in recent research. Providing high system throughput without reducing fairness allocation is becoming a very challenging task. A suitable policy for resource allocation among users is of crucial importance. This study focuses on scheduling multiple streaming flows on the downlink of a WiMAX system based on orthogonal frequency division multiple access (OFDMA). In this paper, we take the first step in formulating and analyzing this problem scrupulously. As a result, we proposed a new scheduling scheme based on Round Robin (RR) Algorithm. Because of its non-opportunistic process, RR does not take in account radio conditions and consequently it affect both system throughput and multi-users diversity. Our contribution called MORRA (Modified Round Robin Opportunistic Algorithm) consists to propose a solution to this issue. MORRA not only exploits the concept of opportunistic scheduler but also takes into account other parameters in the allocation process. The first parameter is called courtesy coefficient (CC) and the second is called Buffer Occupancy (BO). Performance evaluation shows that this well-balanced scheme outperforms both RR and MaxSNR schedulers and demonstrate that choosing between system throughput and fairness is not required.

Keywords: OFDMA, opportunistic scheduling, fairness hierarchy, courtesy coefficient, buffer occupancy

Procedia PDF Downloads 299

Authors: Bahram Khan, Anderson Rocha Ramos, Rui R. Paulo, Fernando J. Velez

Abstract:

With the growing demand for a new blend of applications, the users dependency on the internet is increasing day by day. Mobile internet users are giving more attention to their own experiences, especially in terms of communication reliability, high data rates and service stability on move. This increase in the demand is causing saturation of existing radio frequency bands. To address these challenges, researchers are investigating the best approaches, Carrier Aggregation (CA) is one of the newest innovations, which seems to fulfill the demands of the future spectrum, also CA is one the most important feature for Long Term Evolution - Advanced (LTE-Advanced). For this purpose to get the upcoming International Mobile Telecommunication Advanced (IMT-Advanced) mobile requirements (1 Gb/s peak data rate), the CA scheme is presented by 3GPP, which would sustain a high data rate using widespread frequency bandwidth up to 100 MHz. Technical issues such as aggregation structure, its implementations, deployment scenarios, control signal techniques, and challenges for CA technique in LTE-Advanced, with consideration of backward compatibility, are highlighted in this paper. Also, performance evaluation in macro-cellular scenarios through a simulation approach is presented, which shows the benefits of applying CA, low-complexity multi-band schedulers in service quality, system capacity enhancement and concluded that enhanced multi-band scheduler is less complex than the general multi-band scheduler, which performs better for a cell radius longer than 1800 m (and a PLR threshold of 2%).

Keywords: component carrier, carrier aggregation, LTE-advanced, scheduling

Procedia PDF Downloads 198