Just-in-Time Compilation: The Next Big Thing?
summary
Shows a practical clang-jit (P1609) workflow that turns runtime strings into types, compiles targeted functions on demand, and promotes runtime data into compile-time optimizations. Demonstrations include generating compile-time dispatch from runtime maps, emitting IR from the in-process Clang and then using LLVM to produce a stripped ~1 MB binary (host AOT build was 75 MB because it embeds the compiler). Covers lambdas-as-code, error propagation from Clang, performance gains, and the real risk of C++ injection risk plus mitigations; highlights clang-jit (P1609) and saved IR to produce 1MB binary.
AI-written summary · about
chapters (13)
- 0:00Introduction to JIT
- 1:13Motivation for JIT compilation
- 6:08Existing JIT solutions in C++
- 11:04How Clang JIT works
- 17:28JIT utilities and examples
- 27:25C++ as a scripting language
- 30:16Real-world JIT performance use
- 33:48Compile-time vs runtime dispatching
- 40:16Saving compiled JIT functions
- 44:53JIT and concepts interplay
- 53:17JIT for modding and APIs
- 57:07Summary and future of JIT
- 1:01:21Q&A on JIT vs JavaScript
related talks
01h 01mFaster Compile Times and Better Performance: Bringing Just-in-Time Compilation to C++
26m“Easy::Jit: A Just-in-Time compilation library for C++”
20mLet's Build a Computation-Graph JIT Compiler in C++
30mAn Incremental Approach to JIT Construction
40mJoy of Meta-Tracing Just-in-Time Compilation: More Than Just a VM Generator
39mFireside chat on JIT compiler
19mReusing Just-in-Time Compiled Code
38mFireside chat