Disparity in Semaphores' performance

[sashti srinivasan]

Hello,

     I am writing a performance test-bench for posix
compliant operating systems.  In order to validate my
test-suite, I compared the results with tmtests
distributed with rtems.

     Here is a small set of results:
----------------------------|------------------------
      tmtests               |      my posix tests
----------------------------|-------------------------
 rtems_semaphore_create(9)  |      sem_init (11 usec)
 rtems_semaphore_delete(10) |      sem_destroy(22
usec)
-------------------------------------------------------

     Please guide me whether 12 microseconds
difference between destroying a rtems-semaphore and a
posix-semaphore is theoretically possible.

     In the tmtests, I find calls to functions
Read_timer, Timer_initialize.  I could not find them
in  any document.  Probably they are not
rtems-directives.  I hope Read_timer gives time in
microseconds.  Please correct me if I am wrong.

     In tsim, I find that the function gettimeofday()
returns a same value whenever it is called.  But on
posix bsp, it works.  Any special configuration is
required to get this working on tsim?

Regards
Srinivasan

__________________________________
Do you Yahoo!?
Get better spam protection with Yahoo! Mail.
http://antispam.yahoo.com/tools