Unify benchmark op counters

scripts/compare_baseline.py

@@ -18,12 +18,21 @@ def main() -> int:
     parser.add_argument(
         "--filter-errors",
         action="store_true",
-        help="Only compare rows with err_ops == 0 (default: include all rows)",
+        help="Only compare rows with err_op == 0 (default: include all rows)",
     )
     args = parser.parse_args()
 
     df = pd.read_csv(args.csv_path)
 
+    legacy_columns = {
+        "total_ops": "total_op",
+        "ok_ops": "ok_op",
+        "err_ops": "err_op",
+    }
+    for legacy, current in legacy_columns.items():
+        if legacy in df.columns and current not in df.columns:
+            df = df.rename(columns={legacy: current})
+
     required = {
         "engine",
         "workload_id",
@@ -34,7 +43,7 @@ def main() -> int:
         "read_path",
         "ops",
         "p99_us",
-        "err_ops",
+        "err_op",
     }
     missing = required - set(df.columns)
     if missing:
@@ -50,13 +59,13 @@ def main() -> int:
     ]
 
     if args.filter_errors:
-        base = df[df["err_ops"] == 0].copy()
+        base = df[df["err_op"] == 0].copy()
     else:
         base = df.copy()
 
     if base.empty:
         if args.filter_errors:
-            print("No rows with err_ops == 0, cannot compare.")
+            print("No rows with err_op == 0, cannot compare.")
         else:
             print("No rows found in csv, cannot compare.")
         return 0
@@ -64,13 +73,13 @@ def main() -> int:
     agg = base.groupby(keys + ["engine"], as_index=False).agg(
         ops=("ops", "median"),
         p99_us=("p99_us", "median"),
-        err_ops=("err_ops", "median"),
+        err_op=("err_op", "median"),
     )
 
     piv = agg.pivot_table(
         index=keys,
         columns="engine",
-        values=["ops", "p99_us", "err_ops"],
+        values=["ops", "p99_us", "err_op"],
         aggfunc="first",
     )
     piv.columns = [f"{metric}_{engine}" for metric, engine in piv.columns]
@@ -81,13 +90,13 @@ def main() -> int:
         "ops_rocksdb",
         "p99_us_mace",
         "p99_us_rocksdb",
-        "err_ops_mace",
-        "err_ops_rocksdb",
+        "err_op_mace",
+        "err_op_rocksdb",
     ]:
         if col not in out.columns:
             out[col] = pd.NA
 
-    out["qps_ratio_mace_over_rocksdb"] = (
+    out["ops_ratio_mace_over_rocksdb"] = (
         out["ops_mace"] / out["ops_rocksdb"]
     )
     out["p99_ratio_mace_over_rocksdb"] = out["p99_us_mace"] / out["p99_us_rocksdb"]
@@ -95,7 +104,7 @@ def main() -> int:
 
     print(out.to_string(index=False))
     print("\nInterpretation:")
-    print("- qps_ratio_mace_over_rocksdb > 1: mace has higher throughput")
+    print("- ops_ratio_mace_over_rocksdb > 1: mace has higher throughput")
     print("- p99_ratio_mace_over_rocksdb < 1: mace has lower p99 latency")
 
     return 0