lli

Run LLVM bitcode

$ lli [program.bc]

$ lli [program.ll]

$ lli [program.bc] [arg1] [arg2]

$ lli -force-interpreter=false [program.bc]

$ lli -force-interpreter [program.bc]

$ lli -jit-kind=[orc] [program.bc]

lli [options] file [args]

lli directly executes programs in LLVM bitcode (.bc) or LLVM assembly (.ll) format using a just-in-time (JIT) compiler. Rather than producing a standalone executable on disk, lli compiles the LLVM IR to native machine code in memory at runtime and immediately begins execution, making it a convenient tool for quickly testing and prototyping LLVM IR without a separate compilation and linking step.
The tool supports two execution engines: the JIT compiler (default), which translates functions to native code on demand for near-native performance, and a slower interpreter mode enabled with `-force-interpreter`. The JIT backend can be selected between MCJIT and the newer ORC engine via the `-jit-kind` flag. Program arguments can be passed after the bitcode filename, allowing lli to run LLVM programs just like native executables.

FILE

LLVM bitcode or IR file.

Program arguments.

Use interpreter instead of JIT.

JIT engine type (mcjit, orc).

Entry function name.

Display help information.

Requires LLVM bitcode input. JIT overhead on startup. Part of LLVM toolchain.

lli is part of the LLVM project, providing a way to directly execute LLVM intermediate representation.

llc(1), llvm-as(1), clang(1)