• Share
  • Email
  • Embed
  • Like
  • Save
  • Private Content
No[1][1]
 

No[1][1]

on

  • 705 views

 

Statistics

Views

Total Views
705
Views on SlideShare
705
Embed Views
0

Actions

Likes
0
Downloads
0
Comments
0

0 Embeds 0

No embeds

Accessibility

Categories

Upload Details

Uploaded via as Microsoft PowerPoint

Usage Rights

© All Rights Reserved

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Processing…
Post Comment
Edit your comment

No[1][1] No[1][1] Presentation Transcript

  • 英特网数据中心概述 10/26/2007 James Blakley Director of Server Platform/EPI 优化 , 专用 , 创新
  • 2003 2007E Capital Expenditures (US$ in Millions) Amazon.com Yahoo Google 英特网数据中心的崛起 Source: Morgan Stanley, Thomson, Morningstar 0 1000 2000 3000 4000 5000 6000 MS Live eBay
  • 要成功所面临的挑战
    • 如何支持指数性的不断增长
    • 如何管理大规模的系统安装
    • 如何控制资本和运营的开闸放水
    • 如何在使用环境的限定范围之内有效的实施管理
    • 如何融合新的技术
    • 如何优化处理能力和系统使用率
  • Intel: 一个值得信赖的合作伙伴
    • 服务器诸多元素的不断创新
    • 为 IPDC 客户进行专用设计
    • 优化的处理能力和能耗
    • 不依赖与 OEM 的特定解决方案
  • Intel 与客户的团队式协作 优化 , 专用 , 创新 战略协作关系 软件优化 全新技术的 合作研发 CPU/ 主板 合作 TCO, 能耗管理 数据中心的 优化
  • Backup
  • Customer Case Study: Software Optimization (2006)
    • Background
      • Top IPDC player in the market
      • Optimized key IPDC application running on a very large cluster
      • Delivered best performance/watt solution
      • Used Intel tools to deliver performance (Compilers, VTune, etc.)
      • Source code mods as well as compiler driven optimization
    • Gains
      • Overall +17% gains vs. already well tuned code
        • Profile guided optimizations (+5%)
        • Critical code path function inlining (+5%)
        • Core loop assembly analyzed to schedule better code (+3%)
        • Hand tune to convert indirect calls to direct (+2%)
        • Data layout improved to suit better inlining (+2%)
    • Critical factors for success
      • Visibility to entire source code
      • Access to representative workloads and platforms
      • Ability to use profiling & other performance tools on target platforms
      • Create custom tools for analysis when necessary - e.g. difference of before and after profile data at module, function and assembly level