首页 > 解决方案 > 在 Clojure 中使用 pmap 来并行化归并排序。程序在结束前挂起约 1 分钟,然后终止

问题描述

我正在 Clojure 中编写一个程序,该程序从文本文件 10000.txt (这有 10k 个无符号整数)中获取输入。然后我将该列表传递给我的合并排序函数(一个单线程,2、4、8、16、32线程)。

当我运行程序时,通过键入“clojure test.clj”,它会输出每个函数的经过时间,但程序不会终止。

它挂在那里,就像它在等待输入或即将输出其他东西一样。然后大约 1 分钟后,程序最终终止。幕后一定有什么事情发生。关于它是什么/输入后我需要做什么来终止程序的任何想法?

程序的输出(终止前)这是它挂起大约 1 分钟的地方

终端输出图片

(use 'clojure.java.io)
(require '[clojure.string :as str])

;Read file and store into numbers, as a string
(def numbers (slurp "10000.txt"))

;Parse the string 'numbers', ignore the spaces 
;and save the result into x1 (a lazy-seq of integers)
(def x1 (map #(Integer/parseInt %) (str/split numbers #"\s+")))

;Function that performs the operation of merge sort algorithm
(defn merge-lists [left right]
  (loop [head [] L left R right]
    (if (empty? L) (concat head R)
        (if (empty? R) (concat head L)
            (if (> (first L) (first R))
              (recur (conj head (first R)) L (rest R))
              (recur (conj head (first L)) (rest L) R))))))

;The other merge-sort functions use pmap to run merge sort in parallel
;Using 1,2,4,8,16,32 threads
(defn naive-merge-sort [list]
  (if (< (count list) 2) list
      (apply merge-lists
             (map naive-merge-sort
                  (split-at (/ (count list) 2) list)))))

(defn parallel-merge-sort-2 [list]
  (if (< (count list) 2) list
      (apply merge-lists
             (pmap naive-merge-sort
                   (split-at (/ (count list) 2) list)))))

(defn parallel-merge-sort-4 [list]
  (if (< (count list) 2) list
      (apply merge-lists
             (pmap parallel-merge-sort-2
                   (split-at (/ (count list) 2) list)))))

(defn parallel-merge-sort-8 [list]
  (if (< (count list) 2) list
      (apply merge-lists
             (pmap parallel-merge-sort-4
                   (split-at (/ (count list) 2) list)))))

(defn parallel-merge-sort-16 [list]
  (if (< (count list) 2) list
      (apply merge-lists
             (pmap parallel-merge-sort-8
                   (split-at (/ (count list) 2) list)))))

(defn parallel-merge-sort-32 [list]
  (if (< (count list) 2) list
      (apply merge-lists
             (pmap parallel-merge-sort-16
                   (split-at (/ (count list) 2) list)))))

;Run each of the merge-sort functions and output their time
(time (naive-merge-sort x1))
(time (parallel-merge-sort-2 x1))
(time (parallel-merge-sort-4 x1))
(time (parallel-merge-sort-8 x1))
(time (parallel-merge-sort-16 x1))
(time (parallel-merge-sort-32 x1))

这是我的 10000.txt 文件:https ://pastebin.com/5vKXUk1u

我的预期结果是程序在最后时间打印后终止,而不是 1 分钟终止。

感谢大家的时间和帮助!

标签: multithreadingconcurrencyclojureterminatepmap

解决方案


您需要shutdown-agents在最后调用以停止 Clojure 的线程池。

另请参阅clojure.org 上的代理和异步操作:

请注意,使用代理会启动一个非守护程序后台线程池,这将阻止 JVM 关闭。使用 shutdown-agents 终止这些线程并允许关闭。


推荐阅读