1

2

3

4

5

6

7

$ vmstat 1 100

procs -----------memory---------- ---swap-- -----io---- --system-- -----cpu------

 r  b   swpd   free   buff  cache   si   so    bi    bo   in   cs us sy id wa st

 0  0     88 233484 288756 1784744    0    0     0    23    0    0  4  1 94  0  0

 4  0     88 233236 288756 1784752    0    0     0     0 6202 7880  4  1 96  0  0

 2  0     88 233360 288756 1784800    0    0     0   112 6277 7612  4  1 95  0  0

 0  0     88 232864 288756 1784804    0    0     0   644 5747 6593  6  0 92  2  0

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

linux:~ # pidstat

Linux 2.6.32.12-0.7-default (linux)             06/18/12        _x86_64_

11:37:19          PID    %usr %system  %guest    %CPU   CPU  Command

……

11:37:19        11452    0.00    0.00    0.00    0.00     2  bash

11:37:19        11509    0.00    0.00    0.00    0.00     3  dd

11:37:19: pidstat获取信息时间点

PID: 进程pid

%usr: 进程在用户态运行所占cpu时间比率

%system: 进程在内核态运行所占cpu时间比率

%CPU: 进程运行所占cpu时间比率

CPU: 指示进程在哪个核运行

Command: 拉起进程对应的命令

备注:执行pidstat默认输出信息为系统启动后到执行时间点的统计信息，因而即使当前某进程的cpu占用率很高，输出中的值有可能仍为0。

1

2

3

4

#wget http://www.bitmover.com/lmbench/lmbench3.tar.gz

#tar -zxvf lmbench3.tar.gz

#cd lmbench3

#make

1

2

3

4

5

6

7

8

9

10

11

此时会报错：

   make[2]: *** 没有规则可以创建“bk.ver”需要的目标“../SCCS/s.ChangeSet”。 停止。

   make[2]:正在离开目录 `/home/hero/lmbench3/src'

   make[1]: *** [lmbench] 错误 2

   make[1]:正在离开目录 `/home/hero/lmbench3/src'

   make: *** [build] 错误 2

解决办法：

lmbench3目录下

   #mkdir SCCS

   #touch  ./SCCS/s.ChangeSet

   #make

1

2

3

4

5

6

Context switching - times in microseconds - smaller is better

-------------------------------------------------------------------------

Host                 OS  2p/0K 2p/16K 2p/64K 8p/16K 8p/64K 16p/16K 16p/64K

                         ctxsw  ctxsw  ctxsw ctxsw  ctxsw   ctxsw   ctxsw

--------- ------------- ------ ------ ------ ------ ------ ------- -------

commonway Linux 2.6.18- 9.2400 4.0200 9.0300 7.5600 8.3800    11.6 6.28000

时间的单位是微秒。

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

The benchmark is a ring of two to twenty processes that are connected

with Unix pipes.  A token is passed from process to process, forcing

context switches.  The benchmark measures the time it takes to pass

the token two thousand times from process to process.  Each hand off

of the token has two costs: (a) the context switch, and (b) the cost

of passing the token.  In order to get just the context switching time,the benchmark first measures the cost of passing the token through a

ring of pipes in a single process.  This time is defined as the cost

of passing the token and is not included in the reported context switch

time.

.PP

    When the processes are larger than the default baseline of ``zero''

(where zero means just big enough to do the benchmark), the cost

of the context switch includes the cost of restoring user level

state (cache lines).  This is accomplished by having the process

allocate an array of data and sum it as a series of integers

after receiving the token but before passing the token to the

next process.  Note that the overhead mentioned above includes

the cost of accessing the data but because it is measured in

just one address space, the cost is typically the cost with hot

caches.  So the context switch time does not include anything

other than the context switch provided that all the processes

fit in the cache.  If there are cache misses (as is common), the

cost of the context switch includes the cost of those cache misses.

.PP

    首先是看任务处理的时间（通过一次任务处理，这个任务处理的时间被定义为token时间，不包括线程切换的）。

    然后多次执行，排除任务执行的时间，然后计算出CS的时间（如果有cache miss，则CS的时间也包括cache misses的时间）。