LLI(1)

💅 LLI(1) — the LLVM manual, but make it make sense

okay bestie, welcome to lli(1), part of the LLVM manual. she's a whole vibe and she's about to run your bitcode directly, no middleman, no cap 🫶

✨ NAME ✨

lli — literally just directly execute programs from LLVM bitcode 🦋

✨ SYNOPSIS ✨

lli [options] [filename] [program args]

🌸 DESCRIPTION 🌸

so lli runs programs that are in LLVM bitcode format, like, directly. she takes your program in LLVM bitcode and executes it using a just-in-time compiler or an interpreter. two paths, one icon 👑

real talk though: lli is not an emulator. she will not execute IR for other architectures, and she can only interpret (or JIT-compile) for the host architecture. it's giving "stay in your lane" and honestly? valid.

the JIT compiler takes the same arguments as other tools, like llc, but — lowkey important — they don't necessarily work for the interpreter. 💀

if filename isn't specified, then lli reads the LLVM bitcode for the program from standard input. stdin girlies rise 🙈

the optional args you specify on the command line get passed to the program as arguments. very "here, hold this" of you.

---

🎀 GENERAL OPTIONS 🎀

• -fake-argv0=executable — override the argv[0] value that gets passed into the executing program. yes she can lie about her name, it's fine, it's for science ✨
• -force-interpreter={false,true} — set it to true and she'll use the interpreter even if a just-in-time compiler is available for this architecture. defaults to false.
• -help — prints a summary of command line options. the classic "what does anything do" button 🫶
• -load=pluginfilename — causes lli to load the plugin (shared object) named pluginfilename and use it for optimization. plug her in 💅
• -stats — print statistics from the code-generation passes. heads up: at present this is only meaningful for the just-in-time compiler.
• -time-passes — record how much time each code-generation pass needs and print it to standard error. clocking in ⏱️
• -version — print out the version of lli and exit without doing anything else. she said "here's my ID, bye" 👋

---

🔥 TARGET OPTIONS 🔥

• -mtriple=target triple — override the target triple specified in the input bitcode file with the specified string. ⚠️ this may result in a crash if you pick an architecture that isn't compatible with the current system. so like, choose wisely bestie, no cap.
• -march=arch — specify the architecture to generate assembly for, overriding the target encoded in the bitcode file. see the output of llc -help for a list of valid architectures. by default this is inferred from the target triple or autodetected to the current architecture.
• -mcpu=cpuname — specify a specific chip in the current architecture to generate code for. by default this is inferred from the target triple and autodetected to the current architecture. for a list of available CPUs, use:

llvm-as < /dev/null | llc -march=xyz -mcpu=help

• -mattr=a1,+a2,-a3,... — override or control specific attributes of the target, like whether SIMD operations are enabled or not. the default set of attributes is set by the current CPU. for a list of available attributes, use:

llvm-as < /dev/null | llc -march=xyz -mattr=help

---

🧚 FLOATING POINT OPTIONS 🧚

floats are chaos and these are the knobs, periodt 🍓

• -disable-excess-fp-precision — disable optimizations that may increase floating point precision.
• -enable-no-infs-fp-math — enable optimizations that assume no Inf values.
• -enable-no-nans-fp-math — enable optimizations that assume no NAN values.
• -enable-unsafe-fp-math — causes lli to enable optimizations that may decrease floating point precision. she's being reckless on purpose, don't @ me 😭
• -soft-float — causes lli to generate software floating point library calls instead of equivalent hardware instructions.

---

👑 CODE GENERATION OPTIONS 👑 (the main character section)

• -code-model=model — choose the code model from:

default: Target default code model
tiny: Tiny code model
small: Small code model
kernel: Kernel code model
medium: Medium code model
large: Large code model

• -disable-post-RA-scheduler — disable scheduling after register allocation.
• -disable-spill-fusing — disable fusing of spill code into instructions.
• -jit-enable-eh — exception handling should be enabled in the just-in-time compiler.
• -join-liveintervals — coalesce copies (default=true). she's a joiner 💜
• -nozero-initialized-in-bss — don't place zero-initialized symbols into the BSS section.
• -pre-RA-sched=scheduler — instruction schedulers available (before register allocation):

=default: Best scheduler for the target
=none: No scheduling: breadth first sequencing
=simple: Simple two pass scheduling: minimize critical path and maximize processor utilization
=simple-noitin: Simple two pass scheduling: Same as simple except using generic latency
=list-burr: Bottom-up register reduction list scheduling
=list-tdrr: Top-down register reduction list scheduling
=list-td: Top-down list scheduler

• -regalloc=allocator — register allocator to use (default=linearscan):

=bigblock: Big-block register allocator
=linearscan: linear scan register allocator
=local: local register allocator
=simple: simple register allocator

• -relocation-model=model — choose relocation model from:

=default: Target default relocation model
=static: Non-relocatable code
=pic: Fully relocatable, position independent code
=dynamic-no-pic: Relocatable external references, non-relocatable code

• -spiller — spiller to use (default=local):

=simple: simple spiller
=local: local spiller

• -x86-asm-syntax=syntax — choose style of code to emit from the X86 backend:

=att: Emit AT&T-style assembly
=intel: Emit Intel-style assembly

---

⭐ EXIT STATUS ⭐

if lli fails to load the program, she'll exit with an exit code of 1. otherwise? she returns the exit code of the program she executes. so like, the vibes are inherited 🌈

---

🦋 SEE ALSO 🦋

• llc(1) <llc.html>

---

🌸 Author 🌸

maintained by the LLVM Team (https://llvm.org/). icons, honestly 👑

🎀 Copyright 🎀

2003-2026, LLVM Project

---

and that's the whole page bestie, you ate 💅✨

mwah, this was the LLVM manual page LLI(1), version/source 21, dated 2026-05-15 — go forth and JIT-compile responsibly 🫶💖