Examples of Using prof to Obtain pc Sampling Analysis Data

Examples of Using prof to Obtain pc Sampling Analysis Data

The examples in this section illustrate how to use prof and its options to obtain pc sampling data.

Example Using prof -pcsample

The following partial listing is an example of pc sampling output from a profiled version of the program espresso.

Profile listing generated Fri May 13 14:19:12 1994
    with:       prof -pcsample espresso
-------------------------------------------------------------------------------

samples   time    CPU    FPU   Clock   N-cpu  S-interval Countsize
   4160    42s  R8000  R8010  75.0MHz   1     10.0ms     4(bytes)

Each sample covers 4 bytes for every 10.0ms ( 0.02% of 41.6000sec)

-------------------------------------------------------------------------------
  -p[rocedures] using pc sampling.
  Sorted in descending order by the number of samples in each procedure.
  Unexecuted procedures are excluded.
-------------------------------------------------------------------------------

samples   time(%)      cum time(%)      procedure (file)

    847   8.5s( 20.4)  8.5s( 20.4)  massive_count
(espresso:/usr/people/guest/enjoy/008.espresso/cofactor.c)
    410   4.1s(  9.9)   13s( 30.2)       cofactor
(espresso:/usr/people/guest/enjoy/008.espresso/cofactor.c)
    341   3.4s(  8.2)   16s( 38.4)   setp_implies
(espresso:/usr/people/guest/enjoy/008.espresso/set.c)
    325   3.2s(  7.8)   19s( 46.2)  elim_lowering
(espresso:/usr/people/guest/enjoy/008.espresso/expand.c)
    303     3s(  7.3)   22s( 53.5)    essen_parts
(espresso:/usr/people/guest/enjoy/008.espresso/expand.c)
 ...

In the above listing:

The samples column reports the number of samples in each procedure, sorted in descending order. For example, there were 847 samples for the procedure massive_count.
The time(%) column lists the number of seconds and percentage of execution time spent in each procedure. For example, there were 8.5 seconds (20.4% of execution time) spent in massive_count.
The cum time(%) column lists the percentage of the total execution time spent in each procedure. For example, there were 16 seconds (38.4% of total execution time) were spent cumulatively in the massive_count, cofactor, and setp_implies procedures. Note that this does not imply that these routines called each other; they may have executed sequentially.
The procedure (file) column prints the procedure name and its source file. For example, the source file containing the massive_count procedure is cofactor.c.