Gain a Splash of New Skills - Coursera+ Annual Just ₹7,999
Get 35% Off CFI Certifications - Code CFI35
Overview
Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Explore the diverse landscape of native code compilation technologies in the Java ecosystem through this 55-minute conference talk. Delve into the strengths and weaknesses of Just-In-Time (JIT) and Ahead-Of-Time (AOT) compilers, examining their impact on Java application performance. Discover how JIT compilers profile and speculate on running programs, while AOT compilers generate code before execution. Learn about innovative approaches like cached JIT code, compilation services outside the JVM process, and programmatic native code building. Gain insights into maximizing performance across various Java and JVM-based language applications, understanding the trade-offs between different compilation strategies. Investigate topics such as class hierarchy speculation, profile-directed feedback, and the Open9 Shared Classes Cache, while considering real-world implications for Java developers.
Syllabus
Intro
Native compilers in today's Java ecosystem
Life of a running Java application
1. Speculate on class hierarchy
2. JIT uses profile data collected as program runs
But JIT performance advantage isn't free
BUT there are a few big BUTS
Class loader objects can invalidate the inlining...
Contrived example?
Some AOT Fallacies
Profile Directed Feedback (PDF) may help
Two implementations
Open9 Shared Classes Cache
Strengths and Weaknesses
Wrapping up
Taught by
Java