使用memory_profiler异常
在使用memory_profiler模块0.55.0版本执行命令诊断程序内存用量时,遇到下面错误:
C:\Users\Chen\Desktop\python_doc\第四模块课件>python -m memory_profiler tr1y.py Traceback (most recent call last): File "C:\Users\Chen\Python36\lib\runpy.py", line 193, in _run_module_as_main "__main__", mod_spec) File "C:\Users\Chen\Python36\lib\runpy.py", line 85, in _run_code exec(code, run_globals) File "C:\Users\Chen\Python36\lib\site-packages\memory_profiler.py", line 1227, in <module> exec_with_profiler(script_filename, prof, args.backend, script_args) File "C:\Users\Chen\Python36\lib\site-packages\memory_profiler.py", line 1128, in exec_with_profiler exec(compile(f.read(), filename, 'exec'), ns, ns) UnicodeDecodeError: 'gbk' codec can't decode byte 0x8b in position 331: illegal multibyte sequence
根据提示我找到memory_profiler模块,查看源码line1128附近,原来在line1127出了问题:
因为我用的是国产windows系统,默认字符编码是gbk,若无指定打开文件时也会用gbk,但是我要检测的脚本文件使用utf-8字符编码,因此,改成
with open(filename, encoding='utf-8')就好了,顺利输出诊断信息:
C:\Users\Chen\Desktop\python_doc\第四模块课件>python -m memory_profiler tr1y.py (1.800000000000033, -1.800000000000033) length of x: 1000 total elements: 1000000 Filename: tr1y.py Line # Mem usage Increment Line Contents ================================================ 49 111.680 MiB 111.680 MiB @profile 50 def calculate_z_serial_purepython(max_iter, zs, cs): 51 119.312 MiB 7.633 MiB output = [0] * len(zs) 52 119.336 MiB 0.000 MiB for i in range(len(zs)): 53 119.336 MiB 0.008 MiB n = 0 54 119.336 MiB 0.012 MiB z = zs[i] 55 119.336 MiB 0.000 MiB c = cs[i] 56 119.336 MiB 0.004 MiB while abs(z) < 2 and n < max_iter: 57 119.320 MiB 0.000 MiB z = z * z + c 58 119.320 MiB 0.000 MiB n += 1 59 119.336 MiB 0.000 MiB output[i] = n 60 119.258 MiB 0.000 MiB return output
顺便说一下:从诊断信息来看,在执行line51生成output列表时,可见7MB的RAM被加入这个进程,但这不意味着output列表大小就是7MB,只是进程在列表内部分配时增长了大约7MB。