Fine-Grain Scheduling under Resource Constraints

نویسنده

  • Paul Feautrier
چکیده

Many present-day microprocessors have ne grain parallelism, be it in the form of a pipeline, of multiple functional units, or replicated processors. The eecient use of such architectures depends on the capability of the compiler to schedule the execution of the object code in such a way that most of the available hardware is in use while avoiding so-called dependences. In the case of one simple loop, the schedule may be expressed as an aane form in the loop counter. The coeecient of the loop counter in the schedule is the initiation interval, and gives the mean rate at which loop bodies may be executed. The dependence constraints may be converted to linear inequalities in the coeecients of a closed form schedule, and then solved by classical linear programming algorithms. The resource constraints, however, translate to non-linear constraints. These constraints become linear if the initiation interval is known. This leads to a fast searching algorithm, in which the initiation interval is increased until a feasible solution is found. R esum e Presque tous les microprocesseurs actuels utilisent du parall elisme a grain n, que ce soit sous la forme d'un ou plusieurs pipeline, de multiples unit es fonctionnelles, ou de multiples processeurs. Ces architectures ne peuvent ^ etre utilis ees eecacement que si le compilateur est capable d'organiser l'ex ecution du code objet de faa con a utiliser au mieux le mat eriel disponible tout en respectant les d ependances de donn ees. Dans le cas d'une boucle simple, l'ordonnancement peut ^ etre sp ecii e a l'aide d'une fonction aane du compte-tour, qui donne la date de lancement de chaque it eration de la boucle. Le coeecient du compte tour dans cette fonction est l'intervalle d'initiation de la boucle. Les contraintes de d ependance peuvent ^ etre traduites en contraintes lin eaires sur les coeecients de l'ordonnancement. Le probl eme de minimi-sation de l'intervalle d'initiation peut ^ etre alors r esolu par les algorithmes classiques de la programmation lin eaire. Les contraintes dues a la p enurie de ressources, elles, se traduisent en contraintes non lin eaires, qui deviennent lin eaires d es que l'intervalle d'initiation est connu. Ceci permet de d eenir un algorithme rapide de recherche, a partir d'une borne inf erieure evidente, on accroit l'intervalle d'initiation jusqu'' a trouver une solution.

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

A novel mathematical model for a hybrid flow shop scheduling problem under buffer and resource limitations-A case study

Scheduling problems play a big role in manufacturing and planning the production for increasing the production efficiency and assigning the resources to operations. Furthermore, in many manufacturing systems there is a physical space between stages that called intermediate buffers. In this study, a model is proposed for minimizing the makespan of a hybrid flow shop scheduling problem with inter...

متن کامل

SINGLE MACHINE DUE DATE ASSIGNMENT SCHEDULING PROBLEM WITH PRECEDENCE CONSTRAINTS AND CONTROLLABLE PROCESSING TIMES IN FUZZY ENVIRONMENT

In this paper, a due date assignment scheduling problem with precedence constraints and controllable processing times in uncertain environment is investigated, in which the basic processing time of each job is assumed to be the symmetric trapezoidal fuzzy number, and the linear resource consumption function is used.The objective is to minimize the crisp possibilistic mean (or expected) value of...

متن کامل

Mutation Scheduling: A Unified Approach to Compiling for Fine-Grain Parallelism

Trade-o s between code selection, register allocation, and instruction scheduling are inherently interdependent, especially when compiling for ne-grain parallel architectures. However, the conventional approach to compiling for such machines arbitrarily separates these phases so that decisions made during any one phase place unnecessary constraints on the remaining phases. Mutation Scheduling a...

متن کامل

Mutation Scheduling : A Uni ed Approach toCompiling for Fine - Grain Parallelism ?

Trade-oos between code selection, register allocation, and instruction scheduling are inherently interdependent, especially when compiling for ne-grain parallel architectures. However, the conventional approach to compiling for such machines arbitrarily separates these phases so that decisions made during any one phase place unnecessary constraints on the remaining phases. Mutation Scheduling a...

متن کامل

Tetra: Evaluation of Serial Program Performance on Fine-Grain Parallel Processors

Tetra is a tool for evaluating serial program performance under the resource and control constraints of ne-grain parallel processors. Tetra's primary advantage to the architect is its ability to quickly generate performance metrics for yet to be designed architectures. All the user needs to specify is the capabilities of the architecture (e.g., number of functional units, issue model, etc.), ra...

متن کامل

The Combinatorial Multi-Mode Resource Constrained Multi-Project Scheduling Problem

This paper presents the formulation and solution of the Combinatorial Multi-Mode Resource Constrained Multi-Project Scheduling Problem. The focus of the proposed method is not on finding a single optimal solution, instead on presenting multiple feasible solutions, with cost and duration information to the project manager. The motivation for developing such an approach is due in part to practica...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 1994