Fueling innovative software
July 15-18, 2019
Portland, OR

Performance-tuning Twitter services with Graal and machine learning

Chris Thalinger (Twitter)
2:35pm3:15pm Thursday, July 18, 2019
Secondary topics:  AI Enhanced
Average rating: ****.
(4.60, 5 ratings)

Who is this presentation for?

  • Software engineers




Running Twitter services on Graal has been very successful and saved Twitter a lot of money on data center costs. But Twitter would like to run more efficiently to reduce cost even more—who doesn’t?

Chris Thalinger walks you through how Twitter uses its machine learning framework Autotune to tune Graal inlining parameters and details the performance improvement Twitter showed after autotuning Graal.

Prerequisite knowledge

  • A basic understanding of how Java virtual machines (JVMs) work and what a just-in-time (JIT) compiler is

What you'll learn

  • Learn how Autotune with machine learning can lead to a performance improvement
Photo of Chris Thalinger

Chris Thalinger


Chris Thalinger is a staff software engineer at Twitter who has been working on Java virtual machines for over 14 years. His main expertise is in compiler technology with just-in-time compilation, in particular. He was involved with the CACAO and GNU Classpath projects, but his focus shifted to OpenJDK as soon as Sun made the Java development kit (JDK) open source. Previously, Chris worked on the HotSpot JVM at Sun and Oracle.