Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
Towards Ruby 4 JIT / RubyKaigi 2022
Search
Takashi Kokubun
September 08, 2022
Programming
3
11k
Towards Ruby 4 JIT / RubyKaigi 2022
RubyKaigi 2022
Takashi Kokubun
September 08, 2022
Tweet
Share
More Decks by Takashi Kokubun
See All by Takashi Kokubun
Deoptimization: How YJIT Speeds Up Ruby by Slowing Down / RubyKaigi 2025
k0kubun
2
3.2k
YJIT Makes Rails 1.7x faster / RubyKaigi 2024
k0kubun
7
14k
Ruby JIT Hacking Guide / RubyKaigi 2023
k0kubun
2
10k
YJIT: Dive into Ruby's JIT compiler written in Rust / Rust.Tokyo 2022
k0kubun
2
2.2k
Optimizing Production Performance with MRI JIT / RubyConf 2021
k0kubun
1
470
Why Ruby's JIT was slow / RubyKaigi Takeout 2021
k0kubun
3
2k
数時間かかる週一リリースを毎日何度も爆速でできるようにするまで / CI/CD Conference 2021
k0kubun
21
14k
Ruby 3 JIT's roadmap / RubyConf China 2020
k0kubun
0
830
Ruby 3.0 JIT on Rails
k0kubun
9
9.4k
Other Decks in Programming
See All in Programming
Constant integer division faster than compiler-generated code
herumi
2
670
マイコンでもRustのtestがしたい その2/KernelVM Tokyo 18
tnishinaga
2
2.3k
Claude Code と OpenAI o3 で メタデータ情報を作る
laket
0
130
バイブコーディングの正体——AIエージェントはソフトウェア開発を変えるか?
stakaya
5
980
実践!App Intents対応
yuukiw00w
1
310
パスタの技術
yusukebe
1
390
0から始めるモジュラーモノリス-クリーンなモノリスを目指して
sushi0120
1
310
CLI ツールを Go ライブラリ として再実装する理由 / Why reimplement a CLI tool as a Go library
ktr_0731
3
1.1k
新世界の理解
koriym
0
140
LLMOpsのパフォーマンスを支える技術と現場で実践した改善
po3rin
8
960
Infer入門
riru
4
1.5k
Introduction to Git & GitHub
latte72
0
110
Featured
See All Featured
Done Done
chrislema
185
16k
Unsuck your backbone
ammeep
671
58k
Balancing Empowerment & Direction
lara
2
580
Practical Tips for Bootstrapping Information Extraction Pipelines
honnibal
PRO
23
1.4k
Writing Fast Ruby
sferik
628
62k
RailsConf 2023
tenderlove
30
1.2k
XXLCSS - How to scale CSS and keep your sanity
sugarenia
248
1.3M
Build The Right Thing And Hit Your Dates
maggiecrowley
37
2.8k
Designing for Performance
lara
610
69k
Art, The Web, and Tiny UX
lynnandtonic
302
21k
Statistics for Hackers
jakevdp
799
220k
Design and Strategy: How to Deal with People Who Don’t "Get" Design
morganepeng
131
19k
Transcript
Towards Ruby 4 JIT @k0kubun
@k0kubun Maintain: MJIT, Haml, ERB Shopify team
GitHub Sponsors
Haml 6
Introduction to Ruby JIT
How does Ruby JIT work? Ruby
How does Ruby JIT work? 1 + 2 Ruby Abstract
Syntax Tree
How does Ruby JIT work? 1 + 2 putobject 1
putobject 2 opt_plus leave Ruby Abstract Syntax Tree Instruction Sequence (Bytecode)
How does Ruby JIT work? 1 + 2 putobject 1
putobject 2 opt_plus leave Ruby Abstract Syntax Tree Instruction Sequence (Bytecode) Machine Code
How does Ruby JIT work?
CRuby JIT 1: MJIT
CRuby JIT 2: YJIT
Current CRuby JITs speed.yjit.org
Current CRuby JITs speed.yjit.org
Current CRuby JITs • YJIT • Available since Ruby 3.1
• --jit or --yjit • MJIT • Available since Ruby 2.6 • --mjit
Current CRuby JITs • YJIT • Ruby 3.1: x86_64 only,
no code GC, written in C • Ruby 3.2: arm64 support, (hopefully) code GC, written in Rust • MJIT • Ruby 3.1: Stable-ish, portable, native threads, written in C • Ruby 3.2: Experimental, fork + SIGCHLD, written in Ruby
MJIT in Ruby
None
None
mjit.rb: Secret "standard library" in Ruby 3.2 • mjit.rb •
Even more powerful than TracePoint • You can monkey-patch CRuby JIT • No compatibility guarantee • Every module is private, so const_get is required
BYOJ: Bring Your Own JIT
BYOJ: Bring Your Own JIT • Load and pause MJIT
with --mjit=pause • Define RubyVM::MJIT.compile • Use RubyVM::MJIT.const_get(:C) to hack RubyVM • Call RubyVM::MJIT.resume to start JIT With Ruby 3.2:
YJIT-style JIT • Monkey-patch RubyVM::MJIT.compile
MJIT-style JIT • Monkey-patch RubyVM::MJIT::Compiler.compile
MJIT-style JIT
Everyone is writing CRuby JIT
Benchmarking Ruby JIT
yjit-bench
yjit-bench • yjit-bench has three kinds of benchmarks: 1. Headlining
Benchmarks 2. Other Benchmarks 3. Micro Benchmarks
1. Headlining benchmarks • activerecord • hexapdf • liquid-render •
mail • psych-load • railsbench ✉
2. Other Benchmarks • binarytrees, fankuchredux, nbody • chunky_png •
erubi, erubi_rails • lee • optcarrot • rubykon
3. Micro Benchmarks • 30k_ifelse, 30k_methods • cfunc_itself, str_concat •
fib • getivar, setivar • keyword_args • respond_to
None
Benchmark Your Own JIT • ./run_benchmarks.rb -e “/path/to/ruby --any-option” •
Pass multiple -e options to compare different JITs
Towards Ruby 4 JIT
My wish on Ruby 4 JIT • I want Ruby
4 to be as fast as Java or JavaScript • Ruby 4's performance should be a reason to leave Python
None
More Concrete Examples
None
None
None
Ruby 4 Canary • true is mov-ed (immediate) • No
opt_* VM instruction • Constant folding • Ruby / C method inlining
Ruby 4 Canary’ • Single branch instruction to access @one
• Single register to access two • No heap allocation • No stack frame
None
None
Ruby 4 Canary 2 • 5000050000 is mov-ed (immediate) •
Ruby -> C -> Ruby inlining
How can we get there?
Optimization Challenges 1. Constants 2. Variables 3. Method calls 4.
Garbage collection
1. Constants
1. Constants
1. Constants
1. Constants
1. Constants
1. Constants
1. Constants
1. Constants
2. Variables
2. Variables
2. Variables
2. Variables
2. Variables
2. Variables
2. Variables
2. Variables
2. Variables
2. Variables
2. Variables
2. Variables 2021 2022 (tomorrow)
3. Method calls
3. Method calls
3. Method calls
3. Method calls
3. Method calls
3. Method calls
3. Method calls • Code locality • Method inlining: C
㱻 Ruby • Pass arguments with native ABI • Deoptimization on redefinition or interruption (or TracePoint)
4. Garbage collection
4. Garbage collection
4. Garbage collection
Next Steps • We still have a lot of rooms
for improvements on yjit-bench • More cross-instruction optimizations • More method inlining over Ruby and C
Conclusion • Build your own JIT with Ruby 3.2 •
Benchmark your JIT with yjit-bench