Skip to content

Commit 1d05eb0

Browse files
benghaemmeta-codesync[bot]
authored andcommitted
Switch to to_numeric in normalize_gpu_stream_numbers (#296)
Summary: Pull Request resolved: #296 Saves nearly 2 seconds with equivalent output Reviewed By: A-Kokolis Differential Revision: D82760298 fbshipit-source-id: 254aa99b84f1aefd431f0051433ee0d70b6f347a
1 parent 610b10c commit 1d05eb0

File tree

1 file changed

+2
-11
lines changed

1 file changed

+2
-11
lines changed

hta/utils/utils.py

Lines changed: 2 additions & 11 deletions
Original file line numberDiff line numberDiff line change
@@ -5,7 +5,7 @@
55
import multiprocessing as mp
66
from enum import Enum
77
from pathlib import Path
8-
from typing import Any, List, Tuple
8+
from typing import List, Tuple
99

1010
import pandas as pd
1111
import psutil
@@ -228,16 +228,7 @@ def normalize_gpu_stream_numbers(df: pd.DataFrame) -> None:
228228
logger.error("No stream column found in the trace.")
229229
return
230230

231-
def _normalize_stream_number(stream_number: Any) -> int:
232-
try:
233-
return int(stream_number)
234-
except ValueError:
235-
return -1
236-
237-
df["stream"] = df.apply(
238-
lambda r: _normalize_stream_number(r["stream"]),
239-
axis=1,
240-
)
231+
df["stream"] = pd.to_numeric(df["stream"], errors="coerce").fillna(-1).astype(int)
241232

242233

243234
def get_value_from_dict(d: object, key: str, default: object = None) -> object:

0 commit comments

Comments
 (0)