[ECOS]thread switch frequency

[ECOS]thread switch frequency

[Andrea Acquaviva]

Hi,
I have a question about eCos mlqueue scheduler.
If there are two threads (making a matrix multiplication), running with
the same priority, they switch with a frequency given by:

f_clock/CYGNUM_KERNEL_SCHED_TIMESLICE_TICKS,

where f_clock in the case of strongarm1110 (assabet) is 3.6864MHz.

Is this correct or is there something else that affects the switching
frequency?
My question arises because I observed a mismatch between the switch time
overhead calculated by tm_basic test and the one that I measured by
evaluating the execution time of the two thread mentioned above (that
is, is much less than expected).

Thanks!
Andrea.


--
Ing. Andrea Acquaviva
D.E.I.S. - Universita' di Bologna
V.le Risorgimento, 2    40136 BOLOGNA (ITALY)
Tel: (+39) 051 20 93787 Fax: (+39) 051 2093786

Re: [ECOS]thread switch frequency

[Jonathan Larmour]

Andrea Acquaviva wrote:
> 
> Hi,
> I have a question about eCos mlqueue scheduler.
> If there are two threads (making a matrix multiplication), running with
> the same priority, they switch with a frequency given by:
> 
> f_clock/CYGNUM_KERNEL_SCHED_TIMESLICE_TICKS,
> 
> where f_clock in the case of strongarm1110 (assabet) is 3.6864MHz.
> 
> Is this correct or is there something else that affects the switching
> frequency?

The timer that drives the kernel clock interrupt is at 3.6864MHz yes, but
it is has been programmed to only actually interrupt when its counter
reaches 36864, which corresponds to every 10ms. And only 1 in
CYGNUM_KERNEL_SCHED_TIMESLICE_TICKS of those causes a timeslice.

Jifl
-- 
Red Hat, Rustat House, Clifton Road, Cambridge, UK. Tel: +44 (1223) 271062
Maybe this world is another planet's Hell -Aldous Huxley || Opinions==mine