posts - 319, comments - 22, trackbacks - 0, articles - 11
  C++博客 :: 首页 :: 新随笔 :: 联系 :: 聚合  :: 管理

Mozilla新JavaScript JIT编译器IonMonkey

Posted on 2011-05-04 07:05 RTY 阅读(397) 评论(0)  编辑 收藏 引用 所属分类: 软件

IonMonkey是Mozilla的新JavaScript JIT编译器,旨在为SpiderMonkey JavaScript引擎引入新的优化手段。

InfoQ 采访了IonMonkey首席开发者David Anderson,讨论了其进展,及它为使用SpiderMonkey引擎的产品如FirefoxThunderbird、Adobe Acrobat和MongoDB所带来的性能进步。 新优化方法包括类型特殊化(Type specialization),内联函数,线性扫描寄存器分配算法,无效代码删除,循环不变量移动等。

原文: http://developers.solidot.org/article.pl?sid=11/05/03/1049216&amp

https://wiki.mozilla.org/Platform/Features/IonMonkey

Platform/Features/IonMonkey


Feature Status ETA Owner
IonMonkey Modern JIT Compiler for JavaScript Unknown David Anderson

Summary

IonMonkey is a new JIT for SpiderMonkey. Our goal is to implement a new compiler to enable many new JavaScript optimizations. In particular IonMonkey will feature much more organized and explicit data structures typical of advanced compilers. This goal of being clean and flexible will be extremely important for future optimization work and experimentation.

Team

Who's working on this?

  • Lead Developers: David Anderson, David Mandelin

Release Requirements

Complete checklist of items that need to be satisfied before we can call this feature "done".

  • Does not regress correctness.
  • Does not regress usability or bench-marketing performance.

Next Steps

  • Immediate worklist:
    • Prototype model backend (bug 646923)
    • Design and implement new IRs (bug 650181)
    • Create a code generator and simple register allocator
  • Long-term worklist:
    • Implement method inlining
    • Implement advanced linear-scan register allocation
    • Implement loop invariant code motion
    • Implement constant sub-expression elimination
    • Implement type-profiling
    • Implement enough functionality to replace existing JITs
    • Implement proper memory management of code
    • Implement interval analysis

Designs

See bug 646923 for an experimental design in Python, and bug 650181 for IR plans.

Goals

  • Provide a backend that can match or beat the Trace JIT or Crankshaft in speed. Sub-goals:
    • Fine-grained specialization and de-specialization.
    • Integration with type inference.
  • Clean, textbook IR so optimization passes can be separated and pipelined with well-known algorithms.
  • Document and comment well so the implementation and its side effects can be easily understood.
  • Recompilation, debugging, bailouts are all related - and should be solved up-front.
  • First SpiderMonkey JIT that starts off with peer reviews!
  • (Unknown feasibility) Act as a baseline compiler to replace JM2.
  • Manage memory much better, in part to avoid range problems on x64.

Avoiding these...

  • Aggressive stores. We should move state syncing to bailout points.


Please remove this line and any non-relevant categories below. Add whatever other categories you feel are appropriate.


只有注册用户登录后才能发表评论。
网站导航: 博客园   IT新闻   BlogJava   知识库   博问   管理