Li Benchmarks

nbody_barnes_hut greenvalidity pass

Tier 1 · wall_time · N=N=256,steps=10000 · variant algo_registry · OS linux

passpass — perf claims allowed when ratio is green

Source: latest.csv:perf_present

Wall-clock row exists only — no explicit correctness signal yet.

Ratio vs best competitor: 1.063 (106.3% of julia speed) (table SOTA: li — Li leads; charts use julia = 1.0)
Best competitor value: 2.0154 s

No analytical verify rows in ingest yet. Run tier-1 with --verify so latest.csv exports verify_ulps / verify_within_1ulp.

Problem size: N=N=256,steps=10000 (family nbody_barnes_hut)
Category: micro
Pillar: numerics
Package: lic
Li / catalog oracle (mean ± σ): 2.0154 ± 0.0505 s (n=2) / 2.2147 ± 0.017485 s (n=2) (cpp)
Ratio vs catalog oracle: 0.9100×
Best competitor: li (2.0154 s)
Li relative speed vs SOTA: 1.063 (1.0 = julia speed)
Validity: validity pass
Threshold: 1.2×
Source: lic/benchmarks/workloads/tier2_physics/nbody_barnes_hut

PH ids: PH-5b

Compare oracle: cpp

Relative speed vs best competitor (julia) — SOTA = 1.0, higher is better. Absolute s values are in the table below.

Language comparison
Lang	Mean ± σ	Runs	Unit	Variant	OS
li	2.0154 ± 0.0505	2	s	release	linux
cpp	2.2147 ± 0.017485	2	s	release	linux
rust	2.2108 ± 0.11109	2	s	release	linux
julia	2.1416 ± 0.005607	2	s	release	linux
Metric: wall_time (value = mean of timed runs)