Exploiting thread-levelparallelism (TLP) is a promising way to improve the performance of applications withthe advent of general-purpose cost effective uni-processor and shared-memory multiprocessor systems. In this...
详细信息
ISBN:
(纸本)076951880X
Exploiting thread-levelparallelism (TLP) is a promising way to improve the performance of applications withthe advent of general-purpose cost effective uni-processor and shared-memory multiprocessor systems. In this paper, we describe the OpenMP* implementation in the Intel(R) C++ and Fortran compilers for Intel platforms. We present our major design consideration and decisions in the Intel compiler for generating efficient multithreaded codes guided by OpenMP directives and pragmas. We describe several transformation phases in the compiler for the OpenMP* parallelization. In addition to compiler support, the OpenMP runtime library is a critical part of the Intel compiler. We present runtime techniques developed in the Intel OpenMP runtime library for exploiting thread-levelparallelism as well as integrating the OpenMP support with other forms of threading termed as sibling parallelism. the performance results of a set of benchmarks show good speedups over the well-optimized serial code performance on Intel(R) Pentium- and Itanium-processor based systems.
the proceedings contain 10 papers. the special focus in this conference is on high-levelparallelprogrammingmodels and supportiveenvironments. the topics include: high-level data mapping for clusters of SMPs;integr...
ISBN:
(纸本)3540419446
the proceedings contain 10 papers. the special focus in this conference is on high-levelparallelprogrammingmodels and supportiveenvironments. the topics include: high-level data mapping for clusters of SMPs;integrating task and data parallelism by means of coordination patterns;using loop-levelparallelism to parallelize vectorizable programs;a generic C++ framework for parallel mesh based scientific applications;a portable implementation platform for multithreaded DSM consistency protocols;implementation of a skeleton-based parallelprogramming environment supporting arbitrary nesting;supporting soft real-time tasks and QoS on the java platform;evaluating the XMT parallelprogramming model;a topology-based debugger for MPI programs;correcting errors in message passing systems.
this paper presents the latest version of the SKiPPER skeleton-based parallelprogramming environment dedicated to fast prototyping of vision applications. Compared to the previous version, its main innovative feature...
详细信息
Explicit-multithreading (XMT) is a parallelprogramming model designed for exploiting on-chip parallelism. Its features include a simple thread execution model and an efficient prefix-sum instruction for synchronizing...
详细信息
Following the long tradition of this well established event, the 5thinternationalworkshop on high-levelparallelprogrammingmodelsand supportiveenvironments (HIPS 2000) provides a forum for researchers and develop...
ISBN:
(纸本)354067442X
Following the long tradition of this well established event, the 5thinternationalworkshop on high-levelparallelprogrammingmodelsand supportiveenvironments (HIPS 2000) provides a forum for researchers and developers from both academia and industry to meet and discuss the newest approaches and results in this active research area. It is again held in conjunction with IPDPS (formerly known as IPPS/SPDP), one of the premier events in the area of parallel and distributed processing.
the proceedings contain 10 papers. the topics discussed include: making distributed shared memory simple, yet efficient;constructive and adaptable distributed shared memory;ViC∗: a compiler for virtual-memory C∗ paral...
ISBN:
(纸本)0818684135
the proceedings contain 10 papers. the topics discussed include: making distributed shared memory simple, yet efficient;constructive and adaptable distributed shared memory;ViC∗: a compiler for virtual-memory C∗ parallel and distributed programming with Pthreads and Rthreads;language. bindings for a data-parallel runtime;ZPL's WYSIWYG performance model;a graph-based framework for the definition of tools dealing with sparse and irregular distributed data structures;implementing automatic coordination on networks of workstations;improving performance of multi-dimensional array redistribution on distributed memory machines;and further results for improving loop interchange in non-adjacent and imperfectly nested loops.
In this work we address through a specific example the question about the overhead incurred withthe use of higher abstraction levels for parallelprogramming. We develop a simple molecular dynamics application in ALW...
详细信息
ISBN:
(纸本)0818678836
In this work we address through a specific example the question about the overhead incurred withthe use of higher abstraction levels for parallelprogramming. We develop a simple molecular dynamics application in ALWAN and in MPI, and compare the execution performances on an Intel Paragon machine.
the MILLIPEDE system is a small yet powerful interface of a Virtual parallel Machine (VPM) on top of distributed computing environments. MILLIPEDE is thus a convenient environment for porting various existing parallel...
详细信息
ISBN:
(纸本)0818678836
the MILLIPEDE system is a small yet powerful interface of a Virtual parallel Machine (VPM) on top of distributed computing environments. MILLIPEDE is thus a convenient environment for porting various existing parallelprogramming languages,for the design of new parallelprogramming languages, and for the development of parallel applications. MILLIPEDE is fully implemented at the Technion on a cluster of PCs running Windows-NT.(1) In this paper we briefly, describe the MILLIPEDE interface and discuss the implementation issues of several parallel languages.
暂无评论