Update util.py

delete pynvml and check_gpu_memory

python/paddle_serving_server/env_check/util.py

 import os
-import pynvml
 import argparse
 import base64
 import subprocess
@@ -87,16 +86,6 @@ def kill_process(port, sleep_time=0):
     # 解决端口占用
     os.system(f"sleep {sleep_time}")
 
-
-def check_gpu_memory(gpu_id):
-    pynvml.nvmlInit()
-    handle = pynvml.nvmlDeviceGetHandleByIndex(gpu_id)
-    mem_info = pynvml.nvmlDeviceGetMemoryInfo(handle)
-    mem_used = mem_info.used / 1024 ** 2
-    print(f"GPU-{gpu_id} memory used:", mem_used)
-    return mem_used > 100
-
-
 def count_process_num_on_port(port):
     command = "netstat -nlp | grep :" + str(port) + " | wc -l"
     count = eval(os.popen(command).read())