Re: Realtime Process vs. Time Critical Thread
- From: "Arkady Frenkel" <arkadyf@xxxxxxxxxxxxxxxx>
- Date: Wed, 14 Dec 2005 10:57:05 +0200
Additionally the main players in RT extentions for windows are : RTX of
Ardence ( once VenturCom ) - www.ardence.com , InTime ( RMX ) of TenAsys -
www.tenasys.com and
Kuka ( http://www.kuka.com/germany/en/ )
Arkady
"Arkady Frenkel" <arkadyf@xxxxxxxxxxxxxxxx> wrote in message
news:eUCuSlIAGHA.1268@xxxxxxxxxxxxxxxxxxxxxxx
> In addition to all previous messages :
> If you need to do calculation somehow before and not exactly each 250 ms
> you can do it without problems in XP ( which nicely stand with ten-time
> less than you need periods of time , and the only problem that thread can
> be called somehow around the time needed , that's the problem of general
> purpose OS ). BTW even real time OS
> will not garantee that you'll receive interrupt exactly at the time you
> need but give you opportunities to stride to that. What I mean is in
> modern Windows CE ( 5 , that was from 4 btw ) you have 1 msec timer
> interrupt ( opposite to 10-15 in XP ) but during that time system services
> work to so you'll not receive interrupt exactly 1 msec but around that (
> the same about 250 ). But
> opposite to big windows making your thread priority above normal give your
> thread absolute power over system because all system services run at
> normal priority , that can't be done in XP , where system run on higher
> priority and in the case of time starvation give opportunity to threads
> with low priority to run too.
> Arkady
>
> Arkady
> "JimB" <jayseabee@xxxxxxxxxxxxx> wrote in message
> news:eFZU93$$FHA.2596@xxxxxxxxxxxxxxxxxxxxxxx
>>I am designing a custom robotics application that must compute simple
>>robot trajectories on the fly only 1/4 second ahead of the robot motion. I
>>therefore must guarantee that my thread will wake up every 250mS to do its
>>work. Failure to keep up could result in the machine that is running at 40
>>inches/sec coming to an abrupt stop, or worse, stopping and starting,
>>etc...
>>
>> The application is otherwise a normal priority program consisting mostly
>> of user interface stuff.
>>
>> The computer is running Windows XP Prof, is on a network and connected to
>> a database on another computer, and is otherwise dedicated to running the
>> robotics application.
>>
>> Questions:
>>
>> 1) If I use a single process with a priority class of
>> NORMAL_PRIORITY_CLASS and a worker thread with a thread priority of
>> THREAD_PRIORITY_TIME_CRITICAL can I guarantee in all but the most extreme
>> circumstances that my thread will be responsive enough? Or can I somehow
>> force the worker thread priority to be in the real time range?
>>
>> 2) If I use a single process with a priority class of
>> REALTIME_PRIORITY_CLASS is there a thread priority that can be selected
>> for the main thread that is low enough that the computer will still be
>> responsive then the user interface of my application is responding to
>> normal ui stuff? Or can I somehow force the priority of the main thread
>> to be in the normal priority range?
>>
>>
>
>
.
- References:
- Realtime Process vs. Time Critical Thread
- From: JimB
- Re: Realtime Process vs. Time Critical Thread
- From: Arkady Frenkel
- Realtime Process vs. Time Critical Thread
- Prev by Date: Re: Realtime Process vs. Time Critical Thread
- Next by Date: Re: Thread Issue
- Previous by thread: Re: Realtime Process vs. Time Critical Thread
- Next by thread: Process not always exiting
- Index(es):
Relevant Pages
|
