Understand and Apply Link-Time Optimizations (LTO)

Cross-Unit Optimization at Link

Link-Time Optimization (LTO) compiles to LLVM bitcode in .o files, then runs the full optimization pipeline at link. Clang -flto enables cross-module inlining and devirtualization invisible to separate compilation.

How LTO Works

Each translation unit emits IR bitcode in a special section. The linker invokes the compiler backend as a plugin to merge modules, optimize globally, then codegen once.

bashCopy
Loading…

• Whole-program visibility enables devirtualization
• ThinLTO summarizes modules for faster links
• Debug info is harder across LTO boundaries
• Same IR as compile-time enables identical opts

Connecting to real compilers

Production compilers embed this step inside a longer pipeline. GCC flows through cpp, cc1, assembly, and ld; Clang uses the driver, Sema, LLVM IR passes, and a target backend. LLVM bitcode, JVM bytecode, and WASM are other familiar IRs at the same layer. The exercise isolates one pass so you can test it alone before chaining it to the next stage.

Why for this exercise

You will understand and apply link-time optimization concepts. This exercise asks you to explain how LTO defers codegen and what cross-module wins it enables over separate compilation.