or1ksim: Updated results from RTEMS Tester
heshamelmatary at gmail.com
Sun Sep 21 00:07:42 UTC 2014
On Sun, Sep 21, 2014 at 1:25 AM, Joel Sherrill <joel.sherrill at oarcorp.com>
> On September 20, 2014 6:15:13 PM CDT, Chris Johns <chrisj at rtems.org>
> >On 21/09/2014 8:57 am, Hesham Moustafa wrote:
> >> On Sat, Sep 20, 2014 at 7:53 AM, Chris Johns <chrisj at rtems.org
> >> <mailto:chrisj at rtems.org>> wrote:
> >> On 20/09/2014 3:13 pm, Hesham Moustafa wrote:
> >> I have 4 physical cores, and I usually run make with J8. My
> >> OS is
> >> fedora 20.
> >> Try with --jobs=4 and see if you get any time outs. Anything else
> >> running at the same time may effect the result.
> >> I did, but no big difference, the passed tests only raised to 446
> >> although the system monitor was indicating an average processor
> >> throughput of 75% during this run instance with --job=4.
> >What does --jobs=none return as results ?
> >> It also shows
> >> that I have 8 CPUs (2 virtual ones per one physical core).
> >I am not sure what effect hyper-threading gives us in this case.
> Piling on but just running the long running ones by hand with qemu
> directly and seeing how long they take with time is an option.
ticker is faster on qemu (~8 seconds to finish) rather than or1ksim (for
the same binary). What other tests may take longer? All the tests that
RTEMS Tester tells they timed-out takes no longer that ~30 second for the
longest one when I ran them manually.
Also is this with or without debug enabled? The results are actually pretty
> good for a new port and we want to be diligent. Chris or I need to do a run
> with the same config on sis, psim, and jmr3904 as a comparison.
> I added RTEMS_HEAVY_STACK_DEBUG and RTEMS_HEAVY_MALLOC_DEBUG definitions
at or1ksim/configure.ac to help me debugging the previous bug. These
results were done with there macros still defined there.
> >devel mailing list
> >devel at rtems.org
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the devel