import json
import csv
from pathlib import Path
CRITERION_DIR = Path("target/criterion")
C_CSV = Path("benches/C/flac_bench.csv")
def load_criterion() -> list[dict]:
rows = []
if not CRITERION_DIR.exists():
return rows
for group_dir in CRITERION_DIR.iterdir():
if not group_dir.is_dir():
continue
group = group_dir.name
for bench_dir in group_dir.iterdir():
if not bench_dir.is_dir():
continue
bench_id = bench_dir.name
for case_dir in bench_dir.iterdir():
if not case_dir.is_dir():
continue
case_label = case_dir.name
estimates_path = case_dir / "new" / "estimates.json"
benchmark_path = case_dir / "new" / "benchmark.json"
if not estimates_path.exists():
continue
with open(estimates_path) as f:
est = json.load(f)
throughput_bytes = None
if benchmark_path.exists():
with open(benchmark_path) as f:
bm = json.load(f)
tp = bm.get("throughput")
if tp and isinstance(tp, dict):
throughput_bytes = tp.get("bytes") or tp.get("Bytes")
elif tp and isinstance(tp, list):
for entry in tp:
if isinstance(entry, dict) and ("bytes" in entry or "Bytes" in entry):
throughput_bytes = entry.get("bytes") or entry.get("Bytes")
break
mean_ns = est["mean"]["point_estimate"]
std_ns = est["std_dev"]["point_estimate"]
mbps = None
if throughput_bytes:
mbps = (throughput_bytes / (mean_ns / 1e9)) / 1e6
rows.append({
"source": "rust",
"group": group,
"bench_id": bench_id,
"case_label": case_label,
"mean_ns": mean_ns,
"std_ns": std_ns,
"throughput_bytes": throughput_bytes,
"mbps": mbps,
})
return rows
def load_c() -> list[dict]:
rows = []
if not C_CSV.exists():
return rows
with open(C_CSV) as f:
reader = csv.DictReader(f)
for row in reader:
mbps = None
mean_ns = float(row["mean_ns"])
tb = float(row.get("throughput_bytes") or 0)
if tb > 0 and mean_ns > 0:
mbps = (tb / (mean_ns / 1e9)) / 1e6
rows.append({
"source": "c",
"group": row["group"],
"bench_id": row["bench_id"],
"case_label": row["case_label"],
"mean_ns": mean_ns,
"std_ns": float(row["stdev_ns"]),
"throughput_bytes": tb or None,
"mbps": mbps,
})
return rows
def format_ns(ns: float) -> str:
if ns < 1_000:
return f"{ns:.1f} ns"
if ns < 1_000_000:
return f"{ns/1_000:.1f} µs"
return f"{ns/1_000_000:.1f} ms"
def print_table(title: str, rows: list[dict], group: str):
group_rows = [r for r in rows if r["group"] == group]
if not group_rows:
print(f" (no data for group '{group}')\n")
return
cases = sorted(set(r["case_label"] for r in group_rows))
bench_ids = sorted(set(r["bench_id"] for r in group_rows))
idx = {(r["bench_id"], r["case_label"]): r for r in group_rows}
col_w = 22
id_w = 30
print(f"\n{'─'*10} {title} {'─'*10}")
header = f" {'bench':<{id_w}}" + "".join(f" {c:>{col_w}}" for c in cases)
print(header)
print(f" {'':<{id_w}}" + "".join(f" {'mean (MBps)':>{col_w}}" for _ in cases))
print(" " + "─" * (id_w + (col_w + 2) * len(cases)))
for bid in bench_ids:
cells = []
for case in cases:
r = idx.get((bid, case))
if r is None:
cells.append(f"{'—':>{col_w}}")
elif r["mbps"] is not None:
cells.append(f"{r['mbps']:>{col_w}.1f}")
else:
cells.append(f"{format_ns(r['mean_ns']):>{col_w}}")
print(f" {bid:<{id_w}}" + "".join(f" {c}" for c in cells))
c_libflac = {r["case_label"]: r for r in group_rows if r["bench_id"] == "libFLAC"}
if c_libflac:
print(f"\n {'ratio vs libFLAC (lower=faster)':<{id_w + 2}}")
print(" " + "─" * (id_w + (col_w + 2) * len(cases)))
for bid in bench_ids:
if bid == "libFLAC":
continue
cells = []
for case in cases:
r = idx.get((bid, case))
ref = c_libflac.get(case)
if r and ref and r.get("mbps") and ref.get("mbps"):
ratio = ref["mbps"] / r["mbps"]
marker = " ✓" if ratio <= 1.0 else " "
cells.append(f"{ratio:>{col_w-2}.2f}x{marker}")
else:
cells.append(f"{'—':>{col_w}}")
print(f" {bid:<{id_w}}" + "".join(f" {c}" for c in cells))
print()
def main():
rust_rows = load_criterion()
c_rows = load_c()
all_rows = rust_rows + c_rows
if not all_rows:
print("No benchmark data found.")
print(f" Criterion: {CRITERION_DIR} ({'exists' if CRITERION_DIR.exists() else 'missing'})")
print(f" C CSV: {C_CSV} ({'exists' if C_CSV.exists() else 'missing'})")
return
print(f"\nLoaded {len(rust_rows)} Criterion entries, {len(c_rows)} C entries")
groups = sorted(set(r["group"] for r in all_rows))
for group in groups:
print_table(group, all_rows, group)
if __name__ == "__main__":
main()
