python多线程

处理百万量级数据时，单线程太慢了，笔者第一次使用多线程，记录一下模版

#多线程task代码模版
import multiprocessing as mp

mp.set_start_method('spawn') #gpu多线程需要添加这一句

def xxx_task_xxx_fn(task_list, samples, xxx):
	pass
def xxx_task_xxx(task_list, worker_num, xxx):
	workload = (len(task_list) + worker_num - 1) / worker_num
    workload = int(workload)
    samples = mp.Manager().dict() #list等，根据任务数据格式
    plist = []
    for k in range(worker_num):
        start, end = k*workload, min((k+1)*workload, len(filenames))
        p = mp.Process(target=xxx_task_xxx_fn, args=(task_list[start: end], samples, xxx)
        p.start()
        plist.append(p)
    for p in plist:
        p.join()