A parallel program model for execution time estimation

نویسندگان

چکیده

Рассматриваются программы, выполняемые на видеокартах общего назначения и представленные в виде “ядер”, не содержащих циклов с неопределенной продолжительностью. Такие ядра могут быть реализованы, например, помощью технологий CUDA или OpenCL. Для оценки времени работы подобных программ предложены модели их работы: от совсем “наивной” до более реалистичных. Все они формулируются как матричные выражения max-plus-алгебре. Programs for general-purpose graphics processing units represented as kernels without indefinite loops are considered in this paper. Such can be implemented by or OpenCL technologies, example. For execution time estimation, various models of program introduced: from very “naive” to more reliable. All presented the form matrix expressions max-plus algebra.

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

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Model for the estimation of the execution time of parallel program loops

This paper presents the assumptions and proposition of the model for the estimation of the execution time of parallel program loops. The model is intended to be used in iterative compilation. The model has been elaborated with the focus on shortening the duration of iterative

متن کامل

Parallel Program Execution Anomalies

The parallel runtime depends not only on the utilized algorithm but also on the hardware and software environment properties. Execution time anomalies resulting from unexpected behavior of the hardware and software environment or inexact performance model are well known to every scientist analyzing the performance of the parallel program. An abnormal behavior can usually be observed only for sp...

متن کامل

A Parallel Program Execution Model Supporting Modular Software Construction

A watershed is near in the architecture of computer systems. There is overwhelming demand for systems that support a universal format for computer programs and software components so users may benefit from their use on a wide variety of computing platforms. At present this demand is being met by commodity microprocessors together with standard operating system interfaces. However, current syste...

متن کامل

Visualization of Parallel Program Execution

The paper describes a tool for debugging parallel programs by visualization and animation of their execution behavior. The visualization and animation tool VISTOP (VISualization TOol for Parallel Systems) has originally been developed for a programming library called MMK in a PhD-thesis [Bra94] as part of a tool environment for programming distributed memory multiprocessors. VISTOP supports the...

متن کامل

A model of parallel time estimation

In earlier work, Taatgen, Van Rijn and Anderson (in press) have shown that embedding a simple module that generates temporal information in a more general cognitive architecture explains timing phenomena that were earlier attributed to a hypothesized more complex temporal system. However, the embedded temporal module does not support parallel time estimations, of, for example, two concurrent es...

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

ژورنال

عنوان ژورنال: Vy?islitel?nye metody i programmirovanie

سال: 2022

ISSN: ['1726-3522', '0507-5386']

DOI: https://doi.org/10.26089/nummet.v23r102