sascha/workload-automation
1
0
Fork 0
mirror of https://github.com/ARM-software/workload-automation.git synced 2025-03-22 02:29:10 +00:00
Code Issues Releases Wiki Activity

doc/output: Update output directory information

Browse Source
This commit is contained in:
Marc Bonnici 2018-05-08 17:38:06 +01:00 committed by setrofim
parent 9d0127d685
commit 48a7ad9415
2 changed files with 13 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
doc/source/migration_guide.rst
View File

@ -82,7 +82,7 @@ the new parameter names should be preferred:
searching for a suitable package.
Output ### TODO
Output
^^^^^^^
Output Directory
~~~~~~~~~~~~~~~~

20
doc/source/user_guide.rst
View File

@ -439,24 +439,28 @@ Output Directory
================
The exact contents on the output directory will depend on configuration options
used, instruments and output processors enabled, etc. Typically, the output
directory will contain a results file at the top level that lists all
measurements that were collected (currently, csv and json formats are
supported), along with a subdirectory for each iteration executed with output
for that specific iteration.
used, augmentations enabled, etc.
At the top level, there will also be a ``run.log`` file containing the complete log
At the top level, there will be a ``run.log`` file containing the complete log
output for the execution. The contents of this file is equivalent to what you
would get in the console when using --verbose option.
The output directory will also contain a ``results.json`` file that lists any
overall run metrics and artifacts that have been collected during the run.
Depending on the augmentations that were enabled there may be other results
files in different formats, for example the ``csv``
:ref:`output processor <output-processors>`. In addition you will find a subdirectory
for each successful iteration executed with output and its own ``results.json``
file for that specific iteration.
If a job fails to complete for some reason, then the output directory for that
job will be moved into a new directory called ``__failed``. If the job was
running on a platform that supports android then WA will take a screen capture
and UI dump from the device.
Finally, there will be a ``__meta`` subdirectory. This will contain a copy of
the agenda file used to run the workloads along with any other configuration
files that were supplied for execution.
the agenda file used to run the workloads along with other configuration
files that were used for the execution of that run.
Uninstall
=========