Gregory CLEMENT wrote:
> 2007/5/23, manoj kumar <[EMAIL PROTECTED]>:
>> Hi,
>>
>> i am runnigthe latency test on i386 platform with 1 GB ram and intel pentium
>> 4 runnig at 2.66GHz . i am getting some negative values as lat best and avg.
>> The results are as below
>>
>>
>>     User space interrupt latency output
>>
>>     == Sampling period: 10000 us
>>
>> == Test mode: periodic user-mode task
>>
>> == All results in microseconds
>>
>> warming up...
>>
>> RTT|  00:00:01   (periodic user-mode task, 10000 us period, priority 99)
>>
>> RTH|-----lat min|-----lat avg|-----lat max|-overrun|----lat best|---lat
>> worst
>>
>> RTD|      -1.775|       -1.374|       0.953|       0|      -1.775|
>> 0.953
>>
>> RTD|      -1.790|      -1.355|       3.142|       0|      -1.790|
>> 3.142
>>
>> RTD|      -1.812|       -1.266|       4.158|       0|      -1.812|
>> 4.158
>>
>> RTD|      -1.797|       -1.272|       6.602|       0|      -1.812|
>> 6.602
>>
>> RTD|      -1.764|       -1.365|       0.575|       0|      -1.812|
>> 6.602
>>
>> RTD|      -1.895|       -1.370|       0.859|       0|      -1.895|
>> 6.602
>>
>> RTD|      -1.857|       -1.421|       0.908|       0|      -1.895|
>> 6.602
>>
>> RTD|      -1.782|       -1.293|       5.845|       0|      -1.895|
>> 6.602
>>
>> RTD|      -1.929|       -1.362|       1.204|       0|      -1.929|
>> 6.602
>>
>> RTD|      -1.775|       -1.359|       0.844|       0|      -1.929|
>> 6.602
>>
>> RTD|      -1.794|       -1.346|       0.724|       0|      -1.929|
>> 6.602
>>
>> RTD|      -1.756|      -1.318|       1.021|       0|      -1.929|
>> 6.602
>>
>> RTD|      -1.775|       -1.269|       8.064|       0|      -1.929|
>> 8.064
>>
>> RTD|      -1.805|       -1.395|       0.893|       0|      -1.929|
>> 8.064
>>
>> RTD|      -1.779|        1.530|     281.147|       0|      -1.929|
>> 281.147
>>
>> RTD|      -1.959|       -1.362|       1.118|       0|      -1.959|
>> 281.147
>> Could any figure out why these negative values are coming and reason behind
>> that and how i can avoid these.
>>
>> is it a configuration issuse.
> 
> You have to calibrate xenomai by feeding with your average latency
> /proc/xenomai/latency
> 

...or by setting it at compile time in the kernel .config.

Anyway, I would worry more about that +280 us latency. Probably
SMI-related. Please check TROUBLESHOOTING regarding potential reasons
and solutions.

Jan

_______________________________________________
Xenomai-core mailing list
Xenomai-core@gna.org
https://mail.gna.org/listinfo/xenomai-core

Reply via email to