Rtems #define CONFIGURE_MICROSECONDS_PER_TICK

[Joel Sherrill]

This looks like a benchmark test (tmXX) and they do not include the clock device driver. This you are specifying a parameter that impacts a component not in the program.

FWIW why is the put time call specifying clock tick? It appears to be benchmarking a preempting semaphore release which has no need of a clock tick.

-Joel

Oscar Siby <sibyo at student.chalmers.se> wrote:



Hi,
Im working with a xilinx virtex 5 board running a leon 3 core,my rtems
version is 4.10.0 built from source.
My predicament is that when i change the CONFIGURE_MICROSECONDS_PER_TICK it
does not seem to change anything, here is the code of two of my tasks.

rtems_task High_task(rtems_task_argument argument)
{
  (void) rtems_semaphore_obtain(
    Semaphore_id,
    RTEMS_DEFAULT_OPTIONS,
    0xffffffff
  );
  end_time = benchmark_timer_read();
  put_time(
    "rtems_clock_tick",
    end_time,
    1,
    0,
    CALLING_OVERHEAD_CLOCK_TICK
        );
        rtems_task_delete( RTEMS_SELF );

}

rtems_task Low_task(rtems_task_argument argument)
{
        benchmark_timer_initialize();
        (void) rtems_semaphore_release(Semaphore_id);
        puts( "*** END OF TEST 25 ***" );
        rtems_test_exit( 0 );
}
I have tried with all kinds of values for CONFIGURE_MICROSECONDS_PER_TICK
but the put_time values is still always in the same range. Is the
benchmark_timer running on a different timer/clock ?
/Oscar

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rtems.org/pipermail/users/attachments/20131016/1e2f7416/attachment-0001.html>