Experiment: 5852f59

rad experiment show 5852f59618c833a99a3d75623117cedaf1530daf

by ee@did:key:z6MkvsiybCuk1WDZhh2aXDx7NTiZejKgWZygMNsUuXDMNvVQ · Apr 24 19:32 2026

fat LTO + cgu=1

Measurements

Metric	BaselineThe benchmark measurement of the unmodified code	CandidateThe code with the proposed optimization applied	Candidate/Parent ΔThe performance change from baseline to candidate
duration (ms) primary wall-clock time to compute the content-id of a 64 MiB directory corpus (64 files x 1 MiB each, deterministic PRNG), walking and BLAKE3-hashing every file and then hashing the iroh Collection root; lower is better	8.080 ms (n=1)	7.972 ms (n=1)	-1.33%

Annotations

hypothesis	fat LTO + cgu=1 lets LLVM inline blake3 SIMD compress routines across crate boundaries
what_worked	mild improvement (-1.4%) and lower min-sample — within noise but durable for library consumers using cargo default release profile

Base CommitThe starting commit before the optimization was applied	0494e5ccdd2692ef4b135f1722c0c739833046e7
Candidate CommitThe code with the proposed optimization applied	6d66a81e9e59da5883a9178bf38124d128a79a0e
bench_cmd	`bash ./bench/benchmark.sh`
build_cmd	`cargo build --release -p cc-bench`
Schema	v5

Diff

~ Cargo.toml

@@ -1,6 +1,10 @@
1 [workspace]
2 members = [".", "bench"]
3 
4+[profile.release]
5+lto = "fat"
6+codegen-units = 1
7+
8 [[bin]]
9 name = "rad-artifact"
10 path = "src/bin/rad-artifact.rs"

Environment

Arch	aarch64
OS	Darwin 14.6.1
CPU	Apple M2
Agent	pi-autoresearch / pi
Files	Cargo.toml