projects / apple / shell_cmds.git / blame
| Commit | Line | Data |
|---|---|---|
| 57cab491 A |
1 | .Dd April 24, 2012\r |
| 2 | .Dt systime 1\r | |
| 3 | .Os Darwin\r | |
| 4 | .Sh NAME\r | |
| 5 | .Nm systime\r | |
| 6 | .Nd gather system-wide time during program execution\r | |
| 7 | .Sh SYNOPSIS\r | |
| 8 | .Nm\r | |
| 9 | .Op Fl P\r | |
| 10 | .Ar utility Op Ar argument ...\r | |
| 11 | .Nm\r | |
| 12 | .Fl p\r | |
| 13 | .Nm\r | |
| 14 | .Op Fl P\r | |
| 15 | .Fl u Ar user\r | |
| 16 | .Fl s Ar sys\r | |
| 17 | .Fl i Ar idle\r | |
| 18 | .Sh DESCRIPTION\r | |
| 19 | .Nm\r | |
| 20 | is similar to\r | |
| 21 | .Xr time 1\r | |
| 22 | in that it allows you to execute a utility program. However after execution\r | |
| 23 | completes, it reports the system-wide time that was spent\r | |
| 24 | during that time period, instead of just the portion directly caused\r | |
| 25 | by the utility. This can be used to gather information about CPU\r | |
| 26 | utilization by kernel threads (AIO, networking) and interrupt routines\r | |
| 27 | on behalf of a program, but not directly attributed to it.\r | |
| 28 | .Pp\r | |
| 29 | The\r | |
| 30 | .Nm\r | |
| 31 | program can also be used to snapshot counters with\r | |
| 32 | .Fl p\r | |
| 33 | and then later used with\r | |
| 34 | .Fl u Fl s Fl i\r | |
| 35 | to calculate time spent during a time interval that doesn't correlate\r | |
| 36 | to a specific command execution. For example:\r | |
| 37 | .Pp\r | |
| 38 | .Bd -ragged -offset indent\r | |
| 39 | $ eval `systime -p`\r | |
| 40 | .Pp\r | |
| 41 | $ ... time passes ...\r | |
| 42 | .Pp\r | |
| 43 | $ systime -u $systime_user -s $systime_sys -i $systime_idle\r | |
| 44 | .Pp\r | |
| 45 | 18.79 real 2.03 user 1.04 sys\r | |
| 46 | .Ed\r | |
| 47 | .Pp\r | |
| 48 | .Fl P\r | |
| 49 | can be used to print time spent as a percentage of overall CPU capacity\r | |
| 50 | of the system (capped at 100%)\r | |
| 51 | .Sh SEE ALSO\r | |
| 52 | .Xr time 1 ,\r | |
| 53 | .Xr top 1\r |