并行程序一次要执行很多操作。
正确地运行并行程序是非常困难的。
并行程序易于产生bug不是什么秘密。
我们需要更好的工具去剖析这些并行程序。
We need better tools for profiling such concurrent applications.
想做并行程序，希望大家给点建议！
Want to do parallel programming, I hope we give suggestions!
在编写并行程序时真正困难的事情是使N尽量大。
The real hard work in writing a parallel program is to make N as large as possible.
非正式的并行程序是一个不超过一件事的时间。
Informally, a concurrent program is one that does more than one thing at a time.
与串行程序相比，并行程序调试会遇到新的问题。
Compared with serial programs, parallel programs debugging will face some new problems.
研究表明该算法能减少此类并行程序的完成时间。
It is demonstrated that algorithm can reduce the completion time of the parallel program.
第9章包含了一个关于如何在并行程序中处理错误的详细讨论。
Chapter 9 contains a detailed discussion of how to handle errors in concurrent programs.
介绍了一个跨平台远程并行程序调试器的设计与实现。
This paper introduces the design and implementation of a remote debugger for parallel programs based on multi-platform.
通常的说法是并行程序(和bug)的问题在于它们的不确定性。
It is often said that the problem with concurrent programs (and bugs) is that they are nondeterministic.
ConTest是用于测试、调试和测量并行程序的范围的工具。
ConTest is a tool for testing, debugging, and measuring the coverage of concurrent programs.
并行程序目前是一个活跃的研究领域，也是一个困难问题。
Parallel program is an active research field presently, and it is also a difficult problem.
在MPP环境下的测试表明，改进后的并行程序并行效率高。
It was indicated that modified parallel program had high parallel efficiency under MPP environment.
为了简化大规模并行程序的开发，Google和IBM提供了以下资源。
To simplify the development of massively parallel programs Google and IBM have created the following resources.
在该框架下，程序员可以对多核并行程序的部分正确性进行验证。
Programmers can verify the partial correctness of multi-core parallel programs in our framework.
快速条件断点是高性能并行程序调试器中进程控制必备的功能之一。
A fast conditional breakpoint is considered to be an essential process control capability of advanced high-performance parallel debuggers.
数据分布是影响并行程序在分布主存多处理机上执行性能的重要因素。
Data distribution is one of the key factors that affect the performance of programs running on distributed memory multiprocessors.
在接下来的几节中，我们将编写一些并行程序，并在这个集群上运行这些并行程序。
In the following sections, we will write some parallel programs and run them on this cluster.
本文介绍了并行算法，并介绍了如何编写并行程序、构建集群以及对集群进行基准测试。
This article covers parallel algorithms, and shows you how to write parallel programs, set up clusters, and benchmark clusters.
在本文中，我们介绍了ConTest，一种用于测试、调试和测量并行程序范围的工具。
In this article, we introduce ConTest, a tool for testing, debugging, and measuring the coverage of concurrent programs.
消息传递模型作为一种处理单元间标准信息传输方式在实际并行程序中广为采用。
Message passing model has widely been adapted as a standard information transmission between processing elements in practical parallel programs.
采用对等模式和标准通信模式，在原有的串行程序基础上进行了并行程序的设计。
Parallel programming was designed on the basis of serial programming by using peer-to-peer mode and standard communication mode.
并行程序中的不确定性给任务调度带来了困难，采用动态调度的方法有时开销很大。
The nondeterminism in parallel programs brings many difficulties to the task scheduling in which the dynamic method has high overhead.
作为函数式语言的扩展，并行程序对我来说似乎更自然，尽管我必须承认我没有密切关注其他的。
Parallel programs seem more natural to me as extensions of functional languages; although I must admit that I haven't looked closely for others.
本文主要介绍了BSP模型并通过实例讨论了BSP算法设计和BSP并行程序方法。
In this paper we describe the BSP model and show how to develop parallel programs in BSP style by an example.
Goroutine是轻量级的并行程序执行路径，与线程，coroutine或者进程类似。
Goroutines are lightweight parallel paths of program execution similar to threads, coroutines, or processes.
第2部分将介绍并行算法的知识，并介绍如何编写并行程序、如何构建集群以及如何进行基准测试。
Part 2 covers parallel algorithms, how to write parallel programs, how to set up clusters, and cluster benchmarking.
而现有的并行程序绝大多数是依赖机器的，因此并行程序的可移植性差且难于预测性能。
While present parallel programs mostly machine dependent, so the transportability of parallel programs are very poor and it is very difficult to predict the program performance.
应用推荐