IRQ latency and context switching on mvme5500 (was Re: RTEMSmvme5500 bsp)

[Kate Feng]

Karel Gardas wrote:

> Kate,
>
> your experiments looks nice and in favour of RTEMS, but do you have any
> numbers comparing latencies between receiving ethernet packet and reading
> data from the socket in an application? Or perhaps simple latency
> meassurements between two RTEMS and two vxWorks BSPs connected by ethernet
> with as simple as possible echo service should be fine too and giving
> clear picture.

Do you mean you need  comparison on the network performance of
the two ports  between RTEMS and vxWorks ?   I  did that  using
"EPICS  catime benchmark software" posted at:
http://www.aps.anl.gov/epics/


The result for the 10/100MHZ port is posted at :
http://www.aps.anl.gov/epics/tech-talk/2005/msg00265.php


The  result  for the  1GHZ port  is posted at :
http://www.aps.anl.gov/epics/tech-talk/2005/msg00469.php


Both of them show slightly higher performance
in RTEMS-mvme5500    as the number of channels increase.
I  experienced 'buffer problems'  on the subnets
of NSLS while testing the 10/100MHZ port of the vxWorks-mvme5500.
Someone else experienced the same  problem on
a different subnet  in England with vxWorks-mvme5500.   So far,
I  did not  experience  any  problem with  RTEMS-mvme5500.
I am very happy with its performance.


Regards,
Kate


>
>
> Thanks,
> Karel
>
> On Wed, 27 Jul 2005, Kate Feng wrote:
>
> >
> >>  Interested, yes, but I really have no time right now
> >> to have a closer look at it -- can you summarize the results?
> >
> > Based on the test  result, it shows that  RTEMS-mvme5500 is
> > more deterministic  and  steadier than vxWorks-mvme5500
> > for the highest priority task.  For  both  the idle  and loaded system,
> > RTEMS-mvme5500 "GUARANTEED"  twice or three times
> > faster response time in a steadier state.     The "worst case" is a
> > critical factor  in consideration of the real-time  system.
> >
> >> MVME5500    Interrupt   Latency    |   Context Switching
> >> _________________________________________________________
> >>                            max      (average)   |     max    (average)
> >>
> >> _________________________________________________________
> >> Idle System:                                      |
> >>
> >> RTEMS          5.04       (3.45)           |    6.80     (0.96)
> >>
> >> vxWorks       12.13       (1.58)          |   18.68     (0.91)
> >>
> >> _________________________________________________________
> >> Loaded System:
> >>
> >> RTEMS          8.17       (3.74)          |   24.88     (1.69)
> >>
> >> vxWorks       22.49       (2.04)         |   40.76     (4.1)
> >> _________________________________________________________
> >>
> >> ** All units are in usec.
> >
> >
>
> --
> Karel Gardas                  kgardas at objectsecurity.com
> ObjectSecurity Ltd.          http://www.objectsecurity.com