 |
» |
|
|
|
|
|
dcpicat(1)
NAME
dcpicat - Prints the contents of a profile file.
SYNOPSIS
dcpicat profile files
DESCRIPTION
Dcpicat reads one or more profile files generated by dcpid(1),
and dumps their contents in ASCII to standard output. The output of dcpicat
will mostly be useful to people debugging DCPI, or to people who want the raw
profile data as input to new analysis tools. Users will get more meaningful
output from other analysis tools such as dcpiprof(1), dcpilist(1) and dcpitopcounts(1).
INTERPRETING THE OUTPUT
For each profile file, dcpicat first prints a header that describes
the profile, and then prints the sample counts associated with each program-counter
value. The following sample output from dcpicat is used to explain
the output format:
name xemacs
image 19970328081403c0c18b
label foo
path /usr/local/bin/xemacs
epoch 199805191724
platform walnut
text_start 0x000120000000
text_size 0x000000800000
event 470352 cycles:63488:10000
event 134705 imiss:4096:10000
0x00012006cbb8 29 27
0x00012006cbd0 10 0
0x00012006cbd4 1 0
0x00012006cbd8 2 0
0x00012006cbe0 3 2
0x00012006cbe8 2 0
...
The various fields in the header should be interpreted as follows:
- name
- This is a profile for xemacs.
- image
- The image identifier for xemacs. The first fourteen digits indicate
the UTC time at which the image was created. (The format of the time
is yyyymmddhhmmss.) The last six digits of this number should
match the last six digits of dcpiscan's output for this image.
- label
- The label assigned to this profile. (This field is optional.)
- path
- A path where the image is likely to be found.
- epoch
- The name of the epoch during which the profile was collected.
- platform
- The name of the machine on which the profile
was collected.
- text_start
- The hexadecimal
starting
address of
the text
section of
the image.
- text_size
- The
hexadecimal
byte-size
of
the
text
section
of
the
image.
- event
- One
such
line
is
present
for
each
event
type
present
in
the
profile.
The
example
profile
given
above
contains
two
events:
cycles,
and
imiss.
Each
event
entry
has
the
form "W
X:Y:Z".
The "W" portion
is
the
sum
of
this
event's
count
across
all
pc
values
recorded
in
this
profile. "X" is
the
event
name. "Y" is
the
sampling
period
used
to
sample
this
event:
i.e.,
one
such
event
occurrence
is
recorded
every "Y" occurrences
of
the
event. "Z" is
10000
times
the
fraction
of
the
time
the
event
was
being
monitored.
(When
multiple
events
are
being
multiplexed
onto
the
same
hardware
counter, "Z" will
be
less
than
10000.)
After the header, the multi-column output lists event counts per pc. If
there are "n" events recorded in the profile, there will be "n+1" columns.
The first column is the pc value in increasing address order. Skipped pc
values have zero counts for all events.
The remaining columns contain the count per event type. The columns are
in the same order as the "event" lines in the profile header. For the example
given above, the second column contains cycle counts and the third column
contains imiss counts. For example, the pc value "0x00012006cbb8" has accumulated "29" cycle
samples and "27" imiss samples.
SEE ALSO
dcpi(1), dcpi2bb(1), dcpi2pix(1), dcpi2ps(1), dcpicalc(1), dcpicc(1), dcpicoverage(1), dcpictl(1), dcpid(1), dcpidiff(1), dcpidis(1), dcpiepoch(1), dcpiflow(1), dcpiflush(1), dcpikdiff(1), dcpilabel(1), dcpildlatency(1), dcpilist(1), dcpiprof(1), dcpiprofileme(1), dcpiquit(1), dcpiscan(1), dcpisource(1), dcpistats(1), dcpisumxct(1), dcpitar(1), dcpitopcounts(1), dcpitopstalls(1), dcpiuninstall(1), dcpiupcalls(1), dcpivarg(1), dcpivcat(1), dcpiversion(1), dcpivlst(1), dcpivprofiler(1), dcpiwhatcg(1), dcpix(1), dcpiformat(4), dcpiexclusions(4)
For more information, see the DCPI project home page http://h30097.www3.hp.com/dcpi.
COPYRIGHT
Copyright 1996-2004, Hewlett-Packard Company.
All rights reserved.
|
|
