- Extending Xen* with Intel® Virtualization Technology
- ENERGY STAR* System Implementation
- Competitive Comparison: Dual-Core Intel® Xeon®: Processor-based Platforms vs. AMD Opteron*
- CMP Implementation in Systems Based on the Intel® Core™ Duo processor
- Software Company Plans for Multi-Core: How Epic Games, Adobe Systems, and IBM use Multi-Core Capability
- How to use all of CPUID for x64 platforms under Microsoft Visual Studio .NET 2005
- Flash and .NET Integration using ASP.NET
- Build and consume an ASP.NET Web service
- Multithreaded .NET Web service clients: threads and responsiveness
- High performance image processing and visualization in .NET client applications: Intel Integrated Performance Primitives (IPP)
Welcome to the Intel® Software Dispatch Subscription Program
Overview: The New Era of “Multi-Everywhere”
As the world looks to multi-core processors for new performance gains in mainstream computing, we’re entering a new era of computing. You could call it the era of “multi-everywhere.” This new era involves running multithreaded applications on multi-core processors that enable single-die multiprocessing. This new era is the next logical step in driving Moore’s Law into another decade.
Yet this next step is not without challenges. One of them is Amdahl’s Law. It says that the speedup possible through parallelism (the concurrent execution of tasks to achieve higher performance) will be limited by the portion of the time spent in the sequential component (or serial component) of an application. In other words, no matter how fast the multithreaded parts of a program run, the speedup of the overall program will be limited by the sequential portion of the computation.
This article focuses on a possible solution for improving the performance and energy efficiency of the sequential component in combination with the parallel components of program execution. The solution is EPI throttling. Using it, we can dynamically vary the energy per instruction (EPI) in different cores according to the amount of parallelism available in an application. The result is an asymmetric multiprocessing (AMP) solution that efficiently addresses parallel and sequential parts of the code. Using AMP with EPI throttling, Intel researchers have demonstrated a 38 percent wall clock (real-time) speedup for a wide range of multithreaded programs with sequential components.
To read more, click link below to subscribe to Intel® Software Dispatch and begin receiving Intel® Software Insight, a quarterly e-zine focused on the topics software-industry leaders care about. Once you fill out the brief subscription form, you will be able to download the pdf and continue reading Improving Multi-Core Architecture Power Efficiency through EPI Throttling and Asymmetric Multiprocessing.
To read complete article, click download below.
![]()
If you're interested in this topic, these articles may be helpful:
![]() | Trying to evaluate implementation of a service-oriented architecture? When you start from existing systems, you can think in terms of crea... |
![]() | Intel® Multi-Core Processor Architecture Development by Intel Corporation In the October 1989 issue of IEEE Spectrum, an... |
![]() | Leveraging C++ business logic in a service-oriented architecture Rogue Wave, a Quovadx Inc. division The benefits of an SOA approach... |
![]() | SOA: the new architecture that leverages the old Kerry Champion, founder and chief technology officer, Westbridge Tec... |
![]() | Modeling in the service-oriented architecture by Roger Smith, developer. Intel Corp. By associating a specifica... |
![]()
Related Jobs:


