llc

Compile LLVM IR to assembly

$ llc [input.ll]

$ llc -o [output.s] [input.ll]

$ llc -filetype=obj [input.ll]

$ llc -march=[x86-64] [input.ll]

$ llc -O[2] [input.ll]

$ llc --version

llc [options] file

llc is the LLVM static compiler that translates LLVM intermediate representation (IR) into native machine code. It accepts input in either LLVM bitcode (.bc) or human-readable LLVM assembly (.ll) format and produces target-specific assembly language or object files. As the backend stage of the LLVM compilation pipeline, llc performs instruction selection, register allocation, and machine-specific optimizations to generate efficient code for the target architecture.
The tool supports a wide range of target architectures including x86, ARM, AArch64, MIPS, RISC-V, and others, selectable via the `-march` and `-mtriple` flags. Optimization levels from `-O0` (no optimization) through `-O3` (aggressive optimization) control how much effort llc spends on code improvement during compilation. Output can be either textual assembly (default) or a relocatable object file when using `-filetype=obj`, making llc useful both for inspecting generated code and for producing linkable artifacts directly.

FILE

LLVM bitcode or IR file.

Output file name.

Output type (asm, obj, null).

Target architecture.

Optimization level (0-3).

Target triple.

Display help information.

Requires LLVM IR input. Target support varies. Part of LLVM toolchain.

llc is part of the LLVM compiler infrastructure project, providing the backend compilation step.

clang(1), opt(1), lli(1), llvm-as(1)