Параллельные вычисления

Modern locking

Modern locking

Most multi-threaded software uses locking. Lock optimization traditionally has aimed to reduce lock contention, that is make the critical regions smaller. In optimized software, this often results in a lot of very small critical regions, protected by many locks. Each critical region does only a little work, before releasing the lock and potentially letting some other CPU access the same data.

Measuring Load Imbalance using the Intel® Vtune™ Amplifier XE

OpenMP on the Intel® Xeon Phi™ coprocessor performs as well as on Intel® Xeon processors. However, the slower clock on the Intel Xeon Phi coprocessor and the sheer number of threads accentuates OpenMP overhead.  In most cases, the problem is either load imbalance or a significant amount of serial execution and is rarely the overhead itself.

Let’s take a look at the following Intel Vtune screenshot.

  • Разработчики
  • Профессорский состав
  • Студенты
  • Сервер
  • Intel® VTune™ Amplifier XE
  • MIC
  • Knights Corner
  • Intel Xeon Phi
  • Intel® Many Integrated Core Architecture
  • Оптимизация
  • Параллельные вычисления
  • Многопоточность
  • HTML5标准与性能之一:WebWorkers

    使用HTML5开发的应用越来越多,很多针对性能提升的新标准也开始逐渐提出,提交到HTML5、EcmaScript等标准中,给HTML5的应用更强大的计算能力。我们分4篇文章分别介绍一下这些针对HTML5性能的新标准:Web Workers, Typed Array, ParallelArray 和 asm.js。

    之一:Web Workers

    Web Workers是一个在前端实现多线程的JS接口,属于W3C的HTML5标准的一部分,现在大多数浏览器包括移动浏览器都支持Web Workers.

  • Разработчики
  • Партнеры
  • Студенты
  • Android*
  • Apple iOS*
  • Linux*
  • Microsoft Windows* 8
  • Tizen*
  • HTML5
  • HTML5
  • JavaScript*
  • Начинающий
  • html5
  • WebWorkers
  • Typed Array
  • ParallelArray
  • asm.js
  • Оптимизация
  • Параллельные вычисления
  • Register for Intel® Software Tools Spring Technical Webinar Presentation "Design and prototype scalable threading using Intel® Advisor XE"

    I will be presenting on May 14th at 11am PDT on the following topic:

    Design and prototype scalable threading using Intel® Advisor XE

    Please register for this presentation using the following link:

    https://www1.gotomeeting.com/register/849275177

    Here is a short abstract of the presentation:

    Check out the Intel® Advisor XE 2013 Update 3..

    Intel® Advisor XE 2013 Update 3 guides developers to add parallelism to their existing C/C++ programs. Using this tool, you can identify where most of the time is spent in your code, which of those locations can actually scale to multi-core and what correctness issues are lurking in those locations. The information provided by this tool can help you decide where to thread your code more judiciously. You can learn more about this tool at the Intel(R) Advisor XE 2013 home page.

    SVD multithreading bug in MKL

    Reference Number : DPD200335246

    Product : Intel® Math Kernel Library (Intel® MKL)

    Version : Intel MKL 11.0 Update 3 and Intel MKL 11.1 beta for OS X*

    Operating Systems affected: Windows* OS, Linux * OS, OS X*

    Problem Description : 

  • Разработчики
  • C/C++
  • Fortran
  • Средний
  • Библиотека Intel® Math Kernel Library
  • Параллельные вычисления
  • The Intel Xeon Phi coprocessor: What is it and why should I care? PART 3: Splitting Hares and Tortoises too

    Part 1 showed how to fit 60+ cores onto a single chip. Part 2 showed how those cores can, theoretically, execute up to 3840 instructions in a single clock cycle1.

    Using Pedigrees in Intel® Cilk™ Plus

    Pedigrees are a new feature implemented in Intel Cilk Plus and currently available in Intel® Composer XE 2013. In this post, I explain what pedigrees are, how they work, and how you can use them in Cilk Plus. Pedigrees are a key component used in the implementation of DotMix, a contributed code for a deterministic parallel random-number generator (DPRNG) discussed in my previous post.
  • Разработчики
  • Профессорский состав
  • Студенты
  • C/C++
  • Продвинутый
  • Средний
  • Intel® Cilk™ Plus
  • Cilk Plus
  • pedigrees
  • DPRNG
  • random number generator
  • Параллельные вычисления
  • Многопоточность
  • Страницы

    Подписаться на Параллельные вычисления