One VM To Rule Them All
summary
A research VM stack demonstrates one runtime executing Java, JavaScript, Ruby, R and C with cross-language inlining and near-native speed: a Java-written Graal compiler plus the Truffle AST specialization framework and an LLVM front end (Sulong / Albion) produce peak results competitive with HotSpot and V8. Benchmarks show pure-Ruby image code 80x faster than CRuby, meta‑tracing paths 800–900x faster, streaming gains of 6–10x, and C performance approaching GCC; managed and unmanaged C modes are supported.
AI-written summary · about
chapters (26)
- 0:00Introduction and research context
- 3:21Speed gaps between languages
- 5:30Truffle framework and optimization goals
- 7:36New virtual machine approach
- 10:15Guest languages and Truffle system
- 13:27Partial evaluation and specialization
- 16:19Context-sensitive profiling
- 19:57Trade-offs in language execution
- 21:46Architecture and supported languages
- 23:15JavaScript performance results
- 24:17C performance and GCC comparison
- 25:21Cross-language inlining benefits
- 28:14Data structure abstraction
- 30:05Multilingual programming advantages
- 31:11Ruby image processing benchmark
- 32:43Avoiding C/C++ for performance
- 33:49Meta-programming optimization
- 34:40Project teams and collaboration
- 35:45Current project status and availability
- 37:29Q&A: Managed vs unmanaged C
- 39:55Q&A: Substrate VM performance
- 41:48Q&A: JEP 243 and experimental flags
- 43:54Q&A: Building custom interpreters
- 45:40Q&A: Importance of R language
- 46:42Q&A: Native extensions and challenges
- 48:23Closing remarks and thanks
related talks
50mOne VM to Rule Them All
52mPolyglot on the JVM with Graal: One VM for Java, node.js, Ruby, R, and C++!
20mGraalVM: Run Programs Faster Anywhere
31mSimplifying and Optimizing Life on the Edge with Graal
59mGraalVM: Run Programs Faster Anywhere
37mGraalVM for Polyglots: Harder, Better, Faster, Stronger?
52mGraalVM: Run Programs Faster Anywhere
47mPolyglot Done Right with GraalVM