如何缩短程序运行时间（Time除了监控程序运行时间还能干这个）

如何缩短程序运行时间（Time除了监控程序运行时间还能干这个）Aug 31 16:51:51 ..... started STAR run Aug 31 16:51:51 ... starting to generate Genome Files Aug 31 16:51:52 ..... processing annotations GTF Aug 31 16:51:53 ... starting to sort Suffix Array. This may take a long time... Aug 31 16:51:54 ... sorting Suffix Array chunks and saving them to disk... Aug 31 16:52:10 ... loading chunks from disk packing SA... Aug 31 16:52:13 ... finished generating suf

time是我们比较常用的一个在bash终端监控程序运行的小工具，如

time sleep 2 real 0m2.003s 整个程序运行耗时，从运行开始到运行结束 user 0m0.002s 程序运行过程中用户占用的cpu时间 sys 0m0.001s 程序运行过程中系统占用的CPU时间

那如果我们想监控程序的运行内存怎么办？我们可以调用系统的time，而不是bash的time，什么区别呢？

type time # 这是bash中的time time 是 shell 关键字

# 这是系统的time which time /usr/bin/time

这时我们通常需要全路径去调用，例如

# /usr/bin/time -v command /usr/bin/time -v STAR --runMode genomeGenerate --runThreadN 10 --genomeDir star_GRCh38 --genomeFastaFiles GRCh38.fa --sjdbGTFfile GRCh38.gtf

返回如下信息：

Aug 31 16:51:51 ..... started STAR run Aug 31 16:51:51 ... starting to generate Genome Files Aug 31 16:51:52 ..... processing annotations GTF Aug 31 16:51:53 ... starting to sort Suffix Array. This may take a long time... Aug 31 16:51:54 ... sorting Suffix Array chunks and saving them to disk... Aug 31 16:52:10 ... loading chunks from disk packing SA... Aug 31 16:52:13 ... finished generating suffix array Aug 31 16:52:13 ... generating Suffix Array index Aug 31 16:52:45 ... completed Suffix Array index Aug 31 16:52:45 ..... inserting junctions into the genome indices Aug 31 16:53:05 ... writing Genome to disk ... Aug 31 16:53:06 ... writing Suffix Array to disk ... Aug 31 16:53:09 ... writing SAindex to disk Aug 31 16:53:21 ..... finished successfully Command being timed: "STAR --runMode genomeGenerate --runThreadN 10 --genomeDir star_GRCh38 --genomeFastaFiles GRCh38.fa --sjdbGTFfile GRCh38.gtf" # 用户占用CPU的时间 User time (seconds): 191.97 # 系统占用CPU的时间 System time (seconds): 11.18 # 用户额CPU使用率；虽然指定了10个线程，但只达到了双核效率，快了2倍 Percent of CPU this job got: 223% # 程序从开始运行到结束的时间，人为感觉到的运行了多久 Elapsed (wall clock) time (h:mm:ss or m:ss): 1:30.93 # 下面5项未做统计，都给予了0 Average shared text size (kbytes): 0 Average unshared data size (kbytes): 0 Average stack size (kbytes): 0 Average total size (kbytes): 0 Average resident set size (kbytes): 0 # 该进程占用的最大物理内存 2.3G Maximum resident set size (kbytes): 2306644 # 数据未读入内存，从硬盘读取后缓存入内存 Major (requiring I/O) page faults: 0 # 已读入内存缓存区的数据 Minor (reclaiming a frame) page faults: 2451338 # 程序自动放弃CPU使用权的次数 Voluntary context switches: 293 # 程序被动放弃CPU使用权的次数 Involuntary context switches: 434 Swaps: 0 # 总的读写字符数 File system inputs: 0 File system outputs: 6246432 Socket messages sent: 0 Socket messages received: 0 Signals delivered: 0 Page size (bytes): 4096 Exit status: 0

有了这个工具，下一步就可以愉快的评估不同的软件、数据量所使用的计算资源多少，进而指导电脑装机了。

References

1. https://stackoverflow.com/questions/60779173/what-does-maximum-resident-set-size-mean
2. https://www.gnu.org/software/time/
3. https://ftp.gnu.org/gnu/time/
4. https://stackoverflow.com/questions/998517/problem-with-the-gnu-time-command-to-measure-memory-usage
5. https://superuser.com/questions/480928/is-there-any-command-like-time-but-for-memory-usage
6. https://stackoverflow.com/questions/131303/how-can-i-measure-the-actual-memory-usage-of-an-application-or-process
7. https://stackoverflow.com/questions/774556/peak-memory-usage-of-a-linux-unix-process/774601#774601
8. https://bugzilla.redhat.com/show_bug.cgi?id=702826
9. https://www.slashroot.in/linux-system-io-monitoring