Replace time.time() to time.perf_counter() for benchmarking. (#6178)

Signed-off-by: Lifu Huang <lifu.hlf@gmail.com>

benchmark/tree_of_thought_deep/bench_other.py

@@ -138,7 +138,7 @@ def main(args):
     # Run requests
     states = [None] * len(questions)
 
-    tic = time.time()
+    tic = time.perf_counter()
     if args.backend != "lmql":
 
         def get_one_answer(i):
@@ -177,7 +177,7 @@ def main(args):
             tasks = [get_one_answer_async(k) for k in bt]
             loop.run_until_complete(asyncio.gather(*tasks))
 
-    latency = time.time() - tic
+    latency = time.perf_counter() - tic
 
     answers_text = []
     for s in states:

benchmark/tree_of_thought_deep/bench_sglang.py

@@ -119,7 +119,7 @@ def main(args):
     backend = select_sglang_backend(args)
 
     # Run requests
-    tic = time.time()
+    tic = time.perf_counter()
     states = tree_search.run_batch(
         arguments,
         temperature=0,
@@ -127,7 +127,7 @@ def main(args):
         num_threads=args.parallel,
         progress_bar=True,
     )
-    latency = time.time() - tic
+    latency = time.perf_counter() - tic
     answers_text = []
     for s in states:
         answers_text.append([x for xs in s.ret_value for x in xs])