YoVDO

The State of Julia in 2021 - JuliaCon Keynote

Offered By: The Julia Programming Language via YouTube

Tags

Julia Courses Linear Algebra Courses Automatic Differentiation Courses Garbage Collection Courses LLVM Courses

Course Description

Overview

Dive into a comprehensive 43-minute conference talk from JuliaCon 2021 featuring Julia language creators Stefan Karpinksi, Viral Shah, Jeff Bezanson, and Keno Fischer as they discuss the state of Julia in 2021. Explore the threading and compiler roadmaps, including recent achievements and future goals. Learn about efforts to produce smaller binaries, optimize array operations, and extend the compiler. Discover new features like AbstractInterpreter and OpaqueClosure, and understand their implications. Get insights into the state of automatic differentiation, linear algebra developments, and the performance of key packages like CSV.jl and DataFrames.jl. The talk concludes with a Q&A session addressing separate compilation, conditional dependencies, and potential interpreter speed improvements.

Syllabus

Welcome!.
Opening of the talk.
Threading roadmap.
Threading: things done, somewhat done and not done.
Compiler roadmap: things done in the past year.
Compiler roadmap: things we still need to do.
Separating LLVM and codegen components to produce smaller binaries.
Removing debug info, metadata and LLVM IR from artifacts.
More advanced array optimizations.
Removing speed bumps in GC behavior.
Users extensions of Julia compiler.
New compiler directions.
How do we make it possible to extend the compiler as naturally as extending the library?.
Composability of compiler transformations.
AbstractInterpreter added in Julia 1.6.
Things make possible by AbstractInterpreter.
Limitations of AbstractInterpreter.
OpaqueClouser.
Compiler plugins.
State of the AD.
Linear Algebra Roadmap.
Libblasttrampoline in Julia 1.7.
A native Julia BLAS?.
The future of sparse matrix capabilities in stdlib.
We need more flexibility in our linear algebra stack.
Packages Reaching 1.0 since January 2020.
What percent of register packages have version Julia 1.0+.
Speed of CSV.jl.
Speed of DataFrames.jl.
Q&A: What is a roadmap for separate compilation?.
Q&A: What are the plans on conditional dependencies and how can the community help with it?.
Q&A: How much faster can we make Julia interpreter and how hard it will be?.


Taught by

The Julia Programming Language

Related Courses

ASP.NET Core New Features
LinkedIn Learning
Databricks Certified Data Engineer Associate Cert Prep: 4 Production Pipelines
LinkedIn Learning
First Look: Java 10 and Java 11
LinkedIn Learning
Getting Started as a Java Developer
LinkedIn Learning
Java Concurrency Troubleshooting: Latency and Throughput
LinkedIn Learning