OpenMP task scheduling strategies to mitigate hardware variability in tightly-coupled shared memory clusters

In questa tesi sono stati apportati due importanti contributi nel campo degli acceleratori embedded many-core. Abbiamo implementato un runtime OpenMP ottimizzato per la gestione del tasking model per sistemi a processori strettamente accoppiati in cluster e poi interconnessi attraverso una network on chip. Ci siamo focalizzati sulla loro scalabilità e sul supporto di task di granularità fine, come è tipico nelle applicazioni embedded. Il secondo contributo di questa tesi è stata proporre una estensione del runtime di OpenMP che cerca di prevedere la manifestazione di errori dati da fenomeni di variability tramite una schedulazione efficiente del carico di lavoro.

Accelerating FPGA-based evolution of wavelet transform filters by optimized task scheduling

Adaptive embedded systems are required in various applications. This work addresses these needs in the area of adaptive image compression in FPGA devices. A simplified version of an evolution strategy is utilized to optimize wavelet filters of a Discrete Wavelet Transform algorithm. We propose an adaptive image compression system in FPGA where optimized memory architecture, parallel processing and optimized task scheduling allow reducing the time of evolution. The proposed solution has been extensively evaluated in terms of the quality of compression as well as the processing time. The proposed architecture reduces the time of evolution by 44% compared to our previous reports while maintaining the quality of compression unchanged with respect to existing implementations. The system is able to find an optimized set of wavelet filters in less than 2 min whenever the input type of data changes.

On-Line Real-Time Service-Oriented Task Scheduling Using TUF

We present our approach to real-time service-oriented scheduling problems with the objective of maximizing the total system utility. Different from the traditional utility accrual scheduling problems that each task is associated with only a single time utility function (TUF), we associate two different TUFs—a profit TUF and a penalty TUF—with each task, to model the real-time services that not only need to reward the early completions but also need to penalize the abortions or deadline misses. The scheduling heuristics we proposed in this paper judiciously accept, schedule, and abort real-time services when necessary to maximize the accrued utility. Our extensive experimental results show that our proposed algorithms can significantly outperform the traditional scheduling algorithms such as the Earliest Deadline First (EDF), the traditional utility accrual (UA) scheduling algorithms, and an earlier scheduling approach based on a similar model.

Graph-based User Scheduling for MIMO LEO-based Satellite Communication Systems

The study of the user scheduling problem in a Low Earth Orbit (LEO) Multi-User MIMO system is the objective of this thesis. With the application of cutting-edge digital beamforming algorithms, a LEO satellite with an antenna array and a large number of antenna elements can provide service to many user terminals (UTs) in full frequency reuse (FFR) schemes. Since the number of UTs on-ground are many more than the transmit antennas on the satellite, user scheduling is necessary. Scheduling can be accomplished by grouping users into different clusters: users within the same cluster are multiplexed and served together via Space Division Multiple Access (SDMA), i.e., digital beamforming or Multi-User MIMO techniques; the different clusters of users are then served on different time slots via Time Division Multiple Access (TDMA). The design of an optimal user grouping strategy is known to be an NP-complete problem which can be solved only through exhaustive search. In this thesis, we provide a graph-based user scheduling and feed space beamforming architecture for the downlink with the aim of reducing user inter-beam interference. The main idea is based on clustering users whose pairwise great-circle distance is as large as possible. First, we create a graph where the users represent the vertices, whereas an edge in the graph between 2 users exists if their great-circle distance is above a certain threshold. In the second step, we develop a low complex greedy user clustering technique and we iteratively search for the maximum clique in the graph, i.e., the largest fully connected subgraph in the graph. Finally, by using the 3 aforementioned power normalization techniques, a Minimum Mean Square Error (MMSE) beamforming matrix is deployed on a cluster basis. The suggested scheduling system is compared with a position-based scheduler, which generates a beam lattice on the ground and randomly selects one user per beam to form a cluster.

Energy resource scheduling considering reactive power management

The reactive power management is an important task in future power systems. The control of reactive power allows the increase of distributed energy resources penetration as well as the optimal operation of distribution networks. Currently, the control of reactive power is only controlled in large power units and in high and very high voltage substations. In this paper a reactive power control in smart grids paradigm is proposed, considering the management of distributed energy resources and of the distribution network by an aggregator namely Virtual Power Player (VPP).

A user-centered interface for scheduling problem definition

In this paper we present a user-centered interface for a scheduling system. The purpose of this interface is to provide graphical and interactive ways of defining a scheduling problem. To create such user interface an evaluation-centered user interaction development method was adopted: the star life cycle. The created prototype comprises the Task Module and the Scheduling Problem Module. The first one allows users to define a sequence of operations, i.e., a task. The second one enables a scheduling problem definition, which consists in a set of tasks. Both modules are equipped with a set of real time validations to assure the correct definition of the necessary data input for the scheduling module of the system. The usability evaluation allowed us to measure the ease of interaction and observe the different forms of interaction provided by each participant, namely the reactions to the real time validation mechanism.

Coordination between mid-term maintenance outage decisions and short-term security-constrained scheduling in smart distribution systems

Distribution systems are the first volunteers experiencing the benefits of smart grids. The smart grid concept impacts the internal legislation and standards in grid-connected and isolated distribution systems. Demand side management, the main feature of smart grids, acquires clear meaning in low voltage distribution systems. In these networks, various coordination procedures are required between domestic, commercial and industrial consumers, producers and the system operator. Obviously, the technical basis for bidirectional communication is the prerequisite of developing such a coordination procedure. The main coordination is required when the operator tries to dispatch the producers according to their own preferences without neglecting its inherent responsibility. Maintenance decisions are first determined by generating companies, and then the operator has to check and probably modify them for final approval. In this paper the generation scheduling from the viewpoint of a distribution system operator (DSO) is formulated. The traditional task of the DSO is securing network reliability and quality. The effectiveness of the proposed method is assessed by applying it to a 6-bus and 9-bus distribution system.

Multi-apprentice learning for meta-heuristics parameter tuning in a multi agent scheduling system

The scheduling problem is considered in complexity theory as a NP-hard combinatorial optimization problem. Meta-heuristics proved to be very useful in the resolution of this class of problems. However, these techniques require parameter tuning which is a very hard task to perform. A Case-based Reasoning module is proposed in order to solve the parameter tuning problem in a Multi-Agent Scheduling System. A computational study is performed in order to evaluate the proposed CBR module performance.

Scheduling in holonic manufacturing systems

This chapter presents some of the issues with holonic manufacturing systems. It starts by presenting the current manufacturing scenario and trends and then provides some background information on the holonic concept and its application to manufacturing. The current limitations and future trends of manufacturing suggest more autonomous and distributed organisations for manufacturing systems; holonic manufacturing systems are proposed as a way to achieve such autonomy and decentralisation. After a brief literature survey a specific research work is presented to handle scheduling in holonic manufacturing systems. This work is based on task and resource holons that cooperate with each other based on a variant of the contract net protocol that allow the propagation of constraints between operations in the execution plan. The chapter ends by presenting some challenges and future opportunities of research.

Supporting intra-task parallelism in real-time multiprocessor systems

Os sistemas de tempo real modernos geram, cada vez mais, cargas computacionais pesadas e dinâmicas, começando-se a tornar pouco expectável que sejam implementados em sistemas uniprocessador. Na verdade, a mudança de sistemas com um único processador para sistemas multi- processador pode ser vista, tanto no domínio geral, como no de sistemas embebidos, como uma forma eficiente, em termos energéticos, de melhorar a performance das aplicações. Simultaneamente, a proliferação das plataformas multi-processador transformaram a programação paralela num tópico de elevado interesse, levando o paralelismo dinâmico a ganhar rapidamente popularidade como um modelo de programação. A ideia, por detrás deste modelo, é encorajar os programadores a exporem todas as oportunidades de paralelismo através da simples indicação de potenciais regiões paralelas dentro das aplicações. Todas estas anotações são encaradas pelo sistema unicamente como sugestões, podendo estas serem ignoradas e substituídas, por construtores sequenciais equivalentes, pela própria linguagem. Assim, o modo como a computação é na realidade subdividida, e mapeada nos vários processadores, é da responsabilidade do compilador e do sistema computacional subjacente. Ao retirar este fardo do programador, a complexidade da programação é consideravelmente reduzida, o que normalmente se traduz num aumento de produtividade. Todavia, se o mecanismo de escalonamento subjacente não for simples e rápido, de modo a manter o overhead geral em níveis reduzidos, os benefícios da geração de um paralelismo com uma granularidade tão fina serão meramente hipotéticos. Nesta perspetiva de escalonamento, os algoritmos que empregam uma política de workstealing são cada vez mais populares, com uma eficiência comprovada em termos de tempo, espaço e necessidades de comunicação. Contudo, estes algoritmos não contemplam restrições temporais, nem outra qualquer forma de atribuição de prioridades às tarefas, o que impossibilita que sejam diretamente aplicados a sistemas de tempo real. Além disso, são tradicionalmente implementados no runtime da linguagem, criando assim um sistema de escalonamento com dois níveis, onde a previsibilidade, essencial a um sistema de tempo real, não pode ser assegurada. Nesta tese, é descrita a forma como a abordagem de work-stealing pode ser resenhada para cumprir os requisitos de tempo real, mantendo, ao mesmo tempo, os seus princípios fundamentais que tão bons resultados têm demonstrado. Muito resumidamente, a única fila de gestão de processos convencional (deque) é substituída por uma fila de deques, ordenada de forma crescente por prioridade das tarefas. De seguida, aplicamos por cima o conhecido algoritmo de escalonamento dinâmico G-EDF, misturamos as regras de ambos, e assim nasce a nossa proposta: o algoritmo de escalonamento RTWS. Tirando partido da modularidade oferecida pelo escalonador do Linux, o RTWS é adicionado como uma nova classe de escalonamento, de forma a avaliar na prática se o algoritmo proposto é viável, ou seja, se garante a eficiência e escalonabilidade desejadas. Modificar o núcleo do Linux é uma tarefa complicada, devido à complexidade das suas funções internas e às fortes interdependências entre os vários subsistemas. Não obstante, um dos objetivos desta tese era ter a certeza que o RTWS é mais do que um conceito interessante. Assim, uma parte significativa deste documento é dedicada à discussão sobre a implementação do RTWS e à exposição de situações problemáticas, muitas delas não consideradas em teoria, como é o caso do desfasamento entre vários mecanismo de sincronização. Os resultados experimentais mostram que o RTWS, em comparação com outro trabalho prático de escalonamento dinâmico de tarefas com restrições temporais, reduz significativamente o overhead de escalonamento através de um controlo de migrações, e mudanças de contexto, eficiente e escalável (pelo menos até 8 CPUs), ao mesmo tempo que alcança um bom balanceamento dinâmico da carga do sistema, até mesmo de uma forma não custosa. Contudo, durante a avaliação realizada foi detetada uma falha na implementação do RTWS, pela forma como facilmente desiste de roubar trabalho, o que origina períodos de inatividade, no CPU em questão, quando a utilização geral do sistema é baixa. Embora o trabalho realizado se tenha focado em manter o custo de escalonamento baixo e em alcançar boa localidade dos dados, a escalonabilidade do sistema nunca foi negligenciada. Na verdade, o algoritmo de escalonamento proposto provou ser bastante robusto, não falhando qualquer meta temporal nas experiências realizadas. Portanto, podemos afirmar que alguma inversão de prioridades, causada pela sub-política de roubo BAS, não compromete os objetivos de escalonabilidade, e até ajuda a reduzir a contenção nas estruturas de dados. Mesmo assim, o RTWS também suporta uma sub-política de roubo determinística: PAS. A avaliação experimental, porém, não ajudou a ter uma noção clara do impacto de uma e de outra. No entanto, de uma maneira geral, podemos concluir que o RTWS é uma solução promissora para um escalonamento eficiente de tarefas paralelas com restrições temporais.

Real-time scheduling of parallel tasks in the Linux Kernel

This paper proposes a global multiprocessor scheduling algorithm for the Linux kernel that combines the global EDF scheduler with a priority-aware work-stealing load balancing scheme, enabling parallel real-time tasks to be executed on more than one processor at a given time instant. We state that some priority inversion may actually be acceptable, provided it helps reduce contention, communication, synchronisation and coordination between parallel threads, while still guaranteeing the expected system’s predictability. Experimental results demonstrate the low scheduling overhead of the proposed approach comparatively to an existing real-time deadline-oriented scheduling class for the Linux kernel.

Supporting real-time parallel task models with work-stealing

Dynamic parallel scheduling using work-stealing has gained popularity in academia and industry for its good performance, ease of implementation and theoretical bounds on space and time. Cores treat their own double-ended queues (deques) as a stack, pushing and popping threads from the bottom, but treat the deque of another randomly selected busy core as a queue, stealing threads only from the top, whenever they are idle. However, this standard approach cannot be directly applied to real-time systems, where the importance of parallelising tasks is increasing due to the limitations of multiprocessor scheduling theory regarding parallelism. Using one deque per core is obviously a source of priority inversion since high priority tasks may eventually be enqueued after lower priority tasks, possibly leading to deadline misses as in this case the lower priority tasks are the candidates when a stealing operation occurs. Our proposal is to replace the single non-priority deque of work-stealing with ordered per-processor priority deques of ready threads. The scheduling algorithm starts with a single deque per-core, but unlike traditional work-stealing, the total number of deques in the system may now exceed the number of processors. Instead of stealing randomly, cores steal from the highest priority deque.

Preemption delay analysis for floating non-preemptive region scheduling

In real-time systems, there are two distinct trends for scheduling task sets on unicore systems: non-preemptive and preemptive scheduling. Non-preemptive scheduling is obviously not subject to any preemption delay but its schedulability may be quite poor, whereas fully preemptive scheduling is subject to preemption delay, but benefits from a higher flexibility in the scheduling decisions. The time-delay involved by task preemptions is a major source of pessimism in the analysis of the task Worst-Case Execution Time (WCET) in real-time systems. Preemptive scheduling policies including non-preemptive regions are a hybrid solution between non-preemptive and fully preemptive scheduling paradigms, which enables to conjugate both world's benefits. In this paper, we exploit the connection between the progression of a task in its operations, and the knowledge of the preemption delays as a function of its progression. The pessimism in the preemption delay estimation is then reduced in comparison to state of the art methods, due to the increase in information available in the analysis.

Preemption-light multiprocessor scheduling of sporadic tasks with high utilisation bound

Known algorithms capable of scheduling implicit-deadline sporadic tasks over identical processors at up to 100% utilisation invariably involve numerous preemptions and migrations. To the challenge of devising a scheduling scheme with as few preemptions and migrations as possible, for a given guaranteed utilisation bound, we respond with the algorithm NPS-F. It is configurable with a parameter, trading off guaranteed schedulable utilisation (up to 100%) vs preemptions. For any possible configuration, NPS-F introduces fewer preemptions than any other known algorithm matching its utilisation bound. A clustered variant of the algorithm, for systems made of multicore chips, eliminates (costly) off-chip task migrations, by dividing processors into disjoint clusters, formed by cores on the same chip (with the cluster size being a parameter). Clusters are independently scheduled (each, using non-clustered NPS-F). The utilisation bound is only moderately affected. We also formulate an important extension (applicable to both clustered and non-clustered NPS-F) which optimises the supply of processing time to executing tasks and makes it more granular. This reduces processing capacity requirements for schedulability without increasing preemptions.

Real-time scheduling with resource sharing on uniform multiprocessors

Consider the problem of scheduling a set of implicit-deadline sporadic tasks to meet all deadlines on a uniform multiprocessor platform where each task may access at most one of |R| shared resources and at most once by each job of that task. The resources have to be accessed in a mutually exclusive manner. We propose an algorithm, GIS-vpr, which offers the guarantee that if a task set is schedulable to meet deadlines by an optimal task assignment scheme that allows a task to migrate only when it accesses or releases a resource, then our algorithm also meets the deadlines with the same restriction on the task migration, if given processors 4 + 6|R| times as fast. The proposed algorithm, by design, limits the number of migrations per job to at most two. To the best of our knowledge, this is the first result for resource sharing on uniform multiprocessors with proven performance guarantee.