High performance computing using MPI and OpenMP on multi-core parallel systems

The rapidly increasing number of cores in modern microprocessors is pushing the current high performance computing (HPC) systems into the petascale and exascale era. The hybrid nature of these systems - distributed memory across nodes and shared memory with non-uniform memory access within each node - poses a challenge to application developers. In this paper, we study a hybrid approach to programming such systems - a combination of two traditional programming models, MPI and OpenMP. We present the performance of standard benchmarks from the multi-zone NAS Parallel Benchmarks and two full applications using this approach on several multi-core based systems including an SGI Altix 4700, an IBM p575+ and an SGI Altix ICE 8200EX. We also present new data locality extensions to OpenMP to better match the hierarchical memory structure of multi-core architectures.     

Employing nested OpenMP for the parallelization of multi-zone computational fluid dynamics applications

In this paper we describe the parallelization of the multi-zone code versions of the NAS Parallel Benchmarks employing multi-level OpenMP parallelism. For our study, we use the NanosCompiler that supports nesting of OpenMP directives and provides clauses to control the grouping of threads, load balancing, and synchronization. We report the benchmark results, compare the timings with those of different hybrid parallelization paradigms (MPI+OpenMP and MLP) and discuss OpenMP implementation issues that affect the performance of multi-level parallel applications.     

Supporting Nested OpenMP Parallelism in the TAU Performance System

Nested OpenMP parallelism allows an application to spawn teams of nested threads. This hierarchical nature of thread creation and usage poses problems for performance measurement tools that must determine thread context to properly maintain per-thread performance data. In this paper we describe the problem and a novel solution for identifying threads uniquely. Our approach has been implemented in the TAU performance system and has been successfully used in profiling and tracing OpenMP applications with nested parallelism. We also describe how extensions to the OpenMP standard can help tool developers uniquely identify threads.     

Performance characteristics of the multi-zone NAS parallel benchmarks

We describe a new suite of computational benchmarks that models applications featuring multiple levels of parallelism. Such parallelism is often available in realistic flow computations on systems of meshes, but had not previously been captured in benchmarks. The new suite, named NPB (NAS parallel benchmarks) multi-zone, is derived from the NPB suite, and involves solving the application benchmarks LU, BT and SP on collections of loosely coupled discretization meshes. The solutions on the meshes are updated independently, but after each time step they exchange boundary value information. This strategy provides relatively easily exploitable coarse-grain parallelism between meshes. Three reference implementations are available: one serial, one hybrid using the message passing interface (MPI) and OpenMP, and another hybrid using a shared memory multi-level programming model (SMP+OpenMP). We examine the effectiveness of hybrid parallelization paradigms in these implementations on four different parallel computers. We also use an empirical formula to investigate the performance characteristics of the hybrid parallel codes.     

适合机群OpenMP系统的制导扩展

OpenMP以其易用性和支持增量并行的特点成为共享存储体系结构的编程标准．机群OpenMP系统在机群上实现了OpenMP计算环境，它将OpenMP的易编程性和机群的可扩展性结合起来，是很有意义的．OpenMP的编程方式主要有循环级和SPMD两种，其中循环级方式易于编程而SPMD方式难于编程．然而在机群OpenMP系统中获得高性能OpenMP程序，必需采用SPMD方式．该文描述了适合机群OpenMP系统的一个简单的OpenMP制导扩展子集(包括数据分布制导、循环调度模式)，并在机群OpenMP系统OpenMP／JIAJIA上进行了实现．应用测试表明，利用这些制导扩展进行编程，既保持循环级方式的易编程性又获得与SPMD方式相当的性能，是有效的编程方式．     

CLOMP: Accurately Characterizing OpenMP Application Overheads

Despite its ease of use, OpenMP has failed to gain widespread use on large scale systems, largely due to its failure to deliver sufficient performance. Our experience indicates that the cost of initiating OpenMP regions is simply too high for the desired OpenMP usage scenario of many applications. In this paper, we introduce CLOMP, a new benchmark to characterize this aspect of OpenMP implementations accurately. CLOMP complements the existing EPCC benchmark suite to provide simple, easy to understand measurements of OpenMP overheads in the context of application usage scenarios. Our results for several OpenMP implementations demonstrate that CLOMP identifies the amount of work required to compensate for the overheads observed with EPCC. We also show that CLOMP also captures limitations for OpenMP parallelization on SMT and NUMA systems. Finally, CLOMPI, our MPI extension of CLOMP, demonstrates which aspects of OpenMP interact poorly with MPI when MPI helper threads cannot run on the NIC.     

OpenMP‐oriented applications for distributed shared memory architectures

The rapid rise of OpenMP as the preferred parallel programming paradigm for small‐to‐medium scale parallelism could slow unless OpenMP can show capabilities for becoming the model‐of‐choice for large scale high‐performance parallel computing in the coming decade. The main stumbling block for the adaptation of OpenMP to distributed shared memory (DSM) machines, which are based on architectures like cc‐NUMA, stems from the lack of capabilities for data placement among processors and threads for achieving data locality. The absence of such a mechanism causes remote memory accesses and inefficient cache memory use, both of which lead to poor performance. This paper presents a simple software programming approach called copy‐inside–copy‐back (CC) that exploits the data privatization mechanism of OpenMP for data placement and replacement. This technique enables one to distribute data manually without taking away control and flexibility from the programmer and is thus an alternative to the automat and implicit approaches. Moreover, the CC approach improves on the OpenMP‐SPMD style of programming that makes the development process of an OpenMP application more structured and simpler. The CC technique was tested and analyzed using the NAS Parallel Benchmarks on SGI Origin 2000 multiprocessor machines. This study shows that OpenMP improves performance of coarse‐grained parallelism, although a fast copy mechanism is essential. Copyright © 2004 John Wiley & Sons, Ltd.     

A Proposal for Error Handling in OpenMP

OpenMP has been focused in performance applied to numerical applications, but when we try to move this focus to other kind of applications, like Web servers, we detect one important lack. In these applications, performance is important, but reliability is even more important, and OpenMP does not have any recovery mechanism. In this paper we present a novel proposal to address this lack. In order to add error handling to OpenMP we propose some extensions to the current OpenMP specification. A directive and a clause are proposed, defining a scope for the error handling (where the error can occur) and specifying a behaviour for handling the specific errors. Some examples of use are presented, and we present also an evaluation showing the impact of this proposal in OpenMP applications. We show that this impact is low enough to consider the proposal worthwhile for OpenMP.     

多核环境下边缘提取并行算法研究

随着处理器由高主频的单核处理器逐步转向片上多核处理器(CMP),计算机并行处理能力不断提升。通过分析GIS串行算法面临的性能瓶颈,利用CMP的优势,采用线程级并行处理栅格数据。针对边缘提取算法,深入分析和比较了MPI、OpenMP等当前主流的并行编程模式,提出了并行性能估计模型。基于OpenMP编程模型分析线程数、调度方式和分块大小对算法并行性能的影响,实现边缘提取最优并行。实验证明,性能评估模型能够准确预测CMP环境下的并行性能,基于OpenMP实现的边缘提取并行算法能够提高图像边缘提取效率。     

Optimizing OpenMP Programs on Software Distributed Shared Memory Systems

This paper describes compiler techniques that can translate standard OpenMP applications into code for distributed computer systems. OpenMP has emerged as an important model and language extension for shared-memory parallel programming. However, despite OpenMP's success on these platforms, it is not currently being used on distributed system. The long-term goal of our project is to quantify the degree to which such a use is possible and develop supporting compiler techniques. Our present compiler techniques translate OpenMP programs into a form suitable for execution on a Software DSM system. We have implemented a compiler that performs this basic translation, and we have studied a number of hand optimizations that improve the baseline performance. Our approach complements related efforts that have proposed language extensions for efficient execution of OpenMP programs on distributed systems. Our results show that, while kernel benchmarks can show high efficiency of OpenMP programs on distributed systems, full applications need careful consideration of shared data access patterns. A naive translation (similar to OpenMP compilers for SMPs) leads to acceptable performance in very few applications only. However, additional optimizations, including access privatization, selective touch, and dynamic scheduling, resulting in 31% average improvement on our benchmarks.     

多核构架下OpenMP多线程应用运行性能的研究

多核平台下,OpenMP线程在核间的动态迁移在一定程度上会导致应用程序性能的下降,如果将线程绑定在固定的核上运行,使其不再迁移,这种方法将有可能提升应用程序性能,达到充分利用多核平台的计算能力的目的。本文将介绍如何使用主流的编译器绑定接口以及Linux内核API的方式实现OpenMP线程与核之间的绑定,使用STREAM Benchmark和NPB在上海超级计算中心的"魔方"超级计算机刀片上测试、比较绑定前后的应用程序的性能。结果证明,使用绑定方案将有可能提升OpenMP应用程序的性能。     

3DyRM: a dynamic roofline model including memory latency information

Modern systems present complex memory hierarchies and heterogeneity among cores and processors. As a consequence, efficient programming is challenging. An easy-to-understand performance model, offering guidelines and information about the behaviour of a code, may be useful to alleviate these issues. In this paper, we present two extensions of the well-known Berkeley Roofline Model. The first of these extensions, the Dynamic Roofline Model (DyRM), takes into consideration the complexities of multicore and heterogeneous systems, offering a more detailed view of the evolution of the execution of a code. The second, the 3DyRM, also adds information about the latency of memory accesses to better represent the behaviour on systems with complex memory hierarchies. A set of tools to obtain and represent the models has been implemented. These tools obtain the needed data from hardware counters, with low overhead. Different views are displayed by the tool that can be used to extract the main features of the code. Results of studying, with these tools, the NAS Parallel Benchmarks for OpenMP on two different systems are presented.     

Nebelung: Execution Environment for Transactional OpenMP

Future generations of Chip Multiprocessors (CMP) will provide dozens or even hundreds of cores inside the chip. Writing applications that benefit from the massive computational power offered by these chips is not going to be an easy task for mainstream programmers who are used to sequential algorithms rather than parallel ones. This paper explores the possibility of using Transactional Memory (TM) in OpenMP, the industrial standard for writing parallel programs on shared-memory architectures, for C, C++ and Fortran. One of the major complexities in writing OpenMP applications is the use of critical regions (locks), atomic regions and barriers to synchronize the execution of parallel activities in threads. TM has been proposed as a mechanism that abstracts some of the complexities associated with concurrent access to shared data while enabling scalable performance. The paper presents a first proof-of-concept implementation of OpenMP with TM. Some language extensions to OpenMP are proposed to express transactions. These extensions are implemented in our source-to-source OpenMP Mercurium compiler and our Software Transactional Memory (STM) runtime system Nebelung that supports the code generated by Mercurium. Hardware Transactional Memory (HTM) or Hardware-assisted STM (HaSTM) are seen as possible paths to make the tandem TM-OpenMP more scalable. In the evaluation section we show the preliminary results. The paper finishes with a set of open issues that still need to be addressed, either in OpenMP or in the hardware/software implementations of TM.     

一种适用于机群OpenMP系统的有效调度算法

OpenMP作为共享存储并行编程标准，以其良好的易用性、支持增量并行等特点成为并行程序设计的主流模型之一．OpenMP标准是针对UMA共享存储结构制定的，其循环调度机制只考虑了负载平衡而无须考虑数据分布．然而在机群OpenMP系统中，数据局部性是影响性能的关键因素．针对OpenMP标准中静态调度策略不适合机群计算的缺点，提出了一个充分体现拥有者计算原则的LBS调度算法，并通过扩展制导的方式在机群OpenMP系统（OpenMP/JIAJIA)上加以实现．测试结果表明，LBS算法对于机群OpenMP系统很有效．     

Parallelization of stochastic bounds for Markov chains on multicore and manycore platforms

The author demonstrates the methodology for parallelizing of finding stochastic bounds for Markov chains on multicore and manycore platforms. The stochastic bounds algorithm for Markov chains with the sparse matrices is investigated, thus needing a lot of irregular memory access. Its parallel implementations should scale across multiple threads and characterize with a high performance and performance portability between multicore and manycore platforms. The presented methods are built on the usage of two parallelization extensions of the C++ language: OpenMP and Cilk Plus. For this two extensions, we use two programming models, namely loop parallelism and task-based parallelism. The numerical experiments show the execution time of the implementations and the scalability on multicore and manycore platforms. This work provides the parallel implementations and at the same time presents an educational example of how computer science problems with irregular memory access can be implemented for high performance using OpenMP and Cilk Plus.     

OpenMP on Networks of Workstations for Software DSMs

This paper describes the implementation of a sizable subset of OpenMP on networks of workstations(NOWs) and the source-to-source OpenMP complier(AutoPar） is used for the JIAJIA home-based shared virtual memory system (SVM).The paper suggests some simple modifications and extensions to the OpenMP standard for the difference between SVM and SMP(symmetric multi processor),at which the OpenMP specification is aimed.The OpenMP translator is based on an automatic paralleization compiler,so it is possible to check the correctness of the semantics of OpenMP programs which is not required in an OpenMP-compliant implementation AutoPar is measured for five applications including both programs from NAS Parallel Benchmarks and real applications on a cluster of eight Pentium Ⅱ PCs connected by a 100 Mbps switched Eternet.The evaluation shows that the parallelization by annotaing OpenMPdirectives is simple and the performance of generatd JIAJIA code is still acceptable on NOWs.     

Productivity and Performance Portability of the OpenMP 3.0 Tasking Concept When Applied to an Engineering Code Written in Fortran 95

The modeling of bevel gear cutting processes requires highly flexible data structures and algorithms. We compare the effort and performance of an OpenMP parallelization employing OpenMP 3.0 tasks with previously applied approaches like nesting parallel sections and stack-based algorithms when parallelizing recursive procedures written in Fortran 95 working on binary tree structures. We take a look at various combinations of recent hardware and Fortran compilers.     

基于“魔方”的不同编程模型下科学应用性能比较

本文基于魔方(曙光5000A)"超并行"系统架构比较了不同并行编程模型下NAS Parallel Benchmark Multi-Zone的性能。结果表明,在单节点共享内存存储架构下OpenMP编程模型能获得更好性能,跨节点采用混合编程模型能获得更好性能,测试结果为百万亿次系统的使用及应用性能的提升起到指导作用。     

Runtime vs. Manual Data Distribution for Architecture-Agnostic Shared-Memory Programming Models

This paper compares data distribution methodologies for scaling the performance of OpenMP on NUMA architectures. We investigate the performance of automatic page placement algorithms implemented in the operating system, runtime algorithms based on dynamic page migration, runtime algorithms based on loop scheduling transformations and manual data distribution. These techniques present the programmer with trade-offs between performance and programming effort. Automatic page placement algorithms are transparent to the programmer, but may compromise memory access locality. Dynamic page migration algorithms are also transparent, but require careful engineering and tuned implementations to be effective. Manual data distribution requires substantial programming effort and architecture-specific extensions to the API, but may localize memory accesses in a nearly optimal manner. Loop scheduling transformations may or may not require intervention from the programmer, but conform better to an architecture-agnostic programming paradigm like OpenMP. We identify the conditions under which runtime data distribution algorithms can optimize memory access locality in OpenMP. We also present two novel runtime data distribution techniques, one based on memory access traces and another based on affinity scheduling of parallel loops. These techniques can be used to effectively replace manual data distribution in regular applications. The results provide a proof of concept that it is possible to scale a portable shared-memory programming model up to more than 100 processors, without modifying the API and without exposing architectural details to the programmer.