CPU% in top

ahmedb72 · 09-11-2010, 09:50 AM

Hi,

In output of a top utility, why summation of CPU% is much greater than 100%?

I have RHEL 5.2 with two processors.

Code:

top - 00:46:11 up  1:14,  4 users,  load average: 3.41, 4.17, 3.13
Tasks: 168 total,   6 running, 162 sleeping,   0 stopped,   0 zombie
Cpu(s): 22.1%us, 71.8%sy,  0.0%ni,  6.1%id,  0.0%wa,  0.0%hi,  0.0%si,  0.0%st
Mem:   2075608k total,  1948420k used,   127188k free,    69192k buffers
Swap:  4194296k total,        0k used,  4194296k free,  1599396k cached

%CPU   PID USER      PR  NI  VIRT  RES  SHR S %MEM    TIME+  COMMAND
  32 10133 oracle    16   0 1169m 113m 108m S  5.6   2:24.27 oracleora11gr2 (LOCAL=NO)
  31 10094 oracle    16   0 1169m 113m 108m S  5.6   2:24.41 oracleora11gr2 (LOCAL=NO)
  27 10151 oracle    16   0 1169m 113m 108m R  5.6   2:22.67 oracleora11gr2 (LOCAL=NO)
  26 10131 oracle    16   0 1169m 113m 108m S  5.6   2:22.91 oracleora11gr2 (LOCAL=NO)
  26 10149 oracle    15   0 1169m 113m 108m R  5.6   2:24.03 oracleora11gr2 (LOCAL=NO)
  25 10155 oracle    16   0 1169m 113m 108m S  5.6   2:21.39 oracleora11gr2 (LOCAL=NO)
  22 10139 oracle    15   0 1169m 113m 108m S  5.6   2:19.27 oracleora11gr2 (LOCAL=NO)

AlucardZero · 09-11-2010, 10:00 AM

Because you have two processors. You get 100% per core, so you would have a max of 200%.

druuna · 09-11-2010, 10:00 AM

Quote:

Originally Posted by ahmedb72

Hi,

In output of a top utility, why summation of CPU% is much greater than 100%?

I have RHEL 5.2 with two processors.

Code:

top - 00:46:11 up  1:14,  4 users,  load average: 3.41, 4.17, 3.13
Tasks: 168 total,   6 running, 162 sleeping,   0 stopped,   0 zombie
Cpu(s): 22.1%us, 71.8%sy,  0.0%ni,  6.1%id,  0.0%wa,  0.0%hi,  0.0%si,  0.0%st
Mem:   2075608k total,  1948420k used,   127188k free,    69192k buffers
Swap:  4194296k total,        0k used,  4194296k free,  1599396k cached

%CPU   PID USER      PR  NI  VIRT  RES  SHR S %MEM    TIME+  COMMAND
  32 10133 oracle    16   0 1169m 113m 108m S  5.6   2:24.27 oracleora11gr2 (LOCAL=NO)
  31 10094 oracle    16   0 1169m 113m 108m S  5.6   2:24.41 oracleora11gr2 (LOCAL=NO)
  27 10151 oracle    16   0 1169m 113m 108m R  5.6   2:22.67 oracleora11gr2 (LOCAL=NO)
  26 10131 oracle    16   0 1169m 113m 108m S  5.6   2:22.91 oracleora11gr2 (LOCAL=NO)
  26 10149 oracle    15   0 1169m 113m 108m R  5.6   2:24.03 oracleora11gr2 (LOCAL=NO)
  25 10155 oracle    16   0 1169m 113m 108m S  5.6   2:21.39 oracleora11gr2 (LOCAL=NO)
  22 10139 oracle    15   0 1169m 113m 108m S  5.6   2:19.27 oracleora11gr2 (LOCAL=NO)

22.1 + 71.8 + 6.1 (idle!) = 100.00 %

What gave you the idea that more then 100% was used. BTW: a bit higher then 100% is possible due to rounding "errors".

AlucardZero · 09-11-2010, 10:05 AM

I think he means the %CPU column, druuna

BTW, ahmedb72, try hitting the '1' key to switch between showing one line per core and the summary line at the top.

druuna · 09-11-2010, 10:10 AM

Hi,

Quote:

Originally Posted by AlucardZero

I think he means the %CPU column, druuna

I guess he does

To "merge" both our replies:

As implied by AlucardZero: the percentage in front of the individual process is seen from one cpu. Adding them all together can reach 100 X number of cpu's (200% in your case).

The line I mention: Cpu(s): 22.1%us, 71.8%sy, 0.0%ni, 6.1%id, 0.0%wa, 0.0%hi, 0.0%si, 0.0%st shows the stats for the combined cpu's and 100 % (or a little higher) is the maximum when added together.

Hope this gets rid of any confusion

ahmedb72 · 09-11-2010, 10:42 AM

Hi AlucardZero,
It seems your answer was right. Thanks a lot.

Hi druuna,
I'm counting the column CPU%