Using XSIM
==========

In this section, we have a series of examples which show some ways of using XSIM
to gain a greater insight into how programs run on XCORE devices.

.. _using_xsim_which_tile:

Which tile is my code running on?
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

Build and execute the example in :ref:`multi_tile`:

.. code-block:: bash

  $ xcc -target=XCORE-200-EXPLORER mapfile.xc main.c
  $ xrun --io a.xe
  Hello from tile 0
  Hello from tile 1

How can we convince ourselves that the output was generated by code running on
the specified tiles? We can use :option:`xsim -t` to determine which tile the
system calls were generated by. Here we use :command:`grep` to simplify and
reduce the trace output by capturing 1 line before each of the relevant output
lines:

.. code-block:: bash

  $ xsim -t a.xe | grep -B 1 "Hello from"
  tile[0]@0- -SI A-.----00040264 (_DoSyscall          +  0) : nop      @11057
  Hello from tile 0
  tile[1]@0- -SI A-.----00040234 (_DoSyscall          +  0) : nop      @11061
  Hello from tile 1

This shows that the system calls were generated by the expected tiles.

.. _using_xsim_fast_printf:

Using XScope during simulation
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

You can use XScope when running code on the simulator. It isn't faster than the
non-Xscope approach, but it allows debug of any XScope-related issues.

Build the example in :ref:`fast_printf`. Execute the application using the
:option:`xsim --xscope` option as follows:

.. code-block:: bash

  $ xsim --xscope "-offline xscope.vcd" a.xe
  Tile 0: Result 0
  Tile 1: Iteration 0 Accumulation: 0
  Tile 1: Iteration 1 Accumulation: 1
  Tile 1: Iteration 2 Accumulation: 3
  Tile 1: Iteration 3 Accumulation: 6
  Tile 1: Iteration 4 Accumulation: 10
  Tile 1: Iteration 5 Accumulation: 15
  Tile 1: Iteration 6 Accumulation: 21
  Tile 1: Iteration 7 Accumulation: 28
  Tile 1: Iteration 8 Accumulation: 36
  Tile 1: Iteration 9 Accumulation: 45
  Tile 0: Result 45

XScope trace output will be placed in :file:`xscope.vcd`.