Self-host vs native benchmarks

Regenerate: ./scripts/bench-self-host.sh && python3 scripts/render-self-host-bench-md.py

Compares inauguration’s owned Rust front (in build on in-cli/src/main.rs) to rustc/Cargo building the same CLI crate: compile wall time, binary size, and cold process startup.

What we measure

Stage	Command	Meaning
Owned front + JIT	`in build --path in-cli/src/main.rs`	Parse/type ~1.9k fns; JIT lowering (may fail on stdlib)
rustc release	`cargo build --release -p inauguration`	Full crate + deps link to `target/release/in`
Native self-artifact	`in build --path … --out /tmp/in`	Target: Mach-O from owned pipeline (blocked today)
Stats	`in compile --path … --target jit --json`	`parsed_function_count`, `call_edge_count`

Latest: `in` self-host

Field	Value
Generated (UTC)	2026-07-05T13:43:04Z
`in` version	in 0.7.4
Host / CPU	Darwin / Apple M5 Pro
Functions parsed	1,879
Functions typed	1,879
Call edges	4,268
`in build` wall ms (avg)	197.0
JIT lowering µs (last run)	36,667
Front stats OK	True
Native `--out`	blocked — in build: native compilation failed (native-lowering-failed)

vs rustc (same shipped binary today)

Metric	`in` owned front (`in build` on `main.rs`)	`rustc` / Cargo release (full crate)
Compile wall avg (ms)	197.0	49950.0
Relative compile time	0.004× wall (in ÷ rustc); in front on this incremental compile	baseline (incremental `touch main.rs`)
Shipped `in` binary	69.17 MiB (72,531,952 B)	69.17 MiB (72,531,952 B)
Binary size ratio (in ÷ rustc)	1.000×	same artifact until native self-link
Cold `--version` startup (ms)	157.35	7.33

Notes: Compares front-only parse/type/JIT attempt on in-cli/src/main.rs vs Cargo linking the whole inauguration crate (not a clean cargo clean build). Binary row is today's single target/release/in. Startup row is process launch, not compiler throughput.

Language coverage

Language	Compile	Execute	Notes
`.in`	✅	✅	JIT
`.rs` (in-cli `main.rs`)	✅	⚠️	front OK; JIT/native blocked on `AtomicPtr` etc.
`.rs` (samples)	✅	✅	native Mach-O

Polyglot compile times (under Benchmarks).

Optimization roadmap

Priority	Item	Status
P0	Named/stdlib types in `native_emit` (`AtomicPtr`, atomics)	blocked native + full JIT self-build
P1	Daemon `in daemon start` — drop cold startup on repeated benches	doc’d
P1	Compile cache — persist lowered native / skip re-link `inrt`	partial (source hash)
P2	Incremental Rust front — re-parse from Core IR cache	parallel parse landed
P2	Bench on release tag — script + JSON in repo; optional deploy refresh	script ready
P3	Fair rustc compare — `cargo clean` cold + `in` full native artifact size	future