Board index » delphi » Re: Real-Time control & Delphi

2004-10-16 08:15:23 PM
delphi185
as, for example, medical - to test physical condition/health of possible cosmonauts and jet-pilots.
(I do not provide the link, 'coz it is all over in Russian)
1) mmsystem.dll functions
2) priority routines (with which you can grab complete control over computer).
 

Andrew Rybenkov writes:
en.wikipedia.org/wiki/RTOS
Windows is not a RTOS, since it does not guarantee a minimum response
time. A RTOS does not mean "fast", either. The minimum response time
can be high.
And does not mean you can not create mission-critical tasks on windows,
as you developed yourself.
--
Leonel
www.techtips.com.br
"Any fool can write code that a computer can understand. Good
programmers write code that humans can understand." - Martin Fowler
 

Yes, you can use WIN32 fibers, giving you control over scheduler. But you do
not get control over response time for interrupt service routines (and all
the things that happen in kernel mode). For hard RT issues this also is
important.
Franz
"Andrew Rybenkov" <XXXX@XXXXX.COM>schrieb im Newsbeitrag
 

in this area.
 

Israel Raj T writes:
deployment is to a Linux box with such a real-time kernel (not sure which
one, I am not involved with that part).
--
Wayne Niddery - Logic Fundamentals, Inc. (www.logicfundamentals.com)
RADBooks: www.logicfundamentals.com/RADBooks.html
Working for yourself is great because you get to work half days, and
you can choose any twelve hours you want.
 

Andrew Rybenkov writes:
the OS that the next frame of your video will be processed in time for
it to be displayed when it should be.
Don't mix the apparent real-time operation of Windows and the
requirements industry are asking for. Windows cannot guarantee that a
given operation performs in at most x milliseconds. 99% of the time, it
does, but you have no warranty. and those 1% may be critical when
controlling external systems
 

"John Herbster" <herb-sci1_AT_sbcglobal.net>writes
tasks if one knows
what one is doing and the task suits. Of course you can. In fact, that's
exactly what I want to do!
But Windows on its own is not classified as an rtos from everything I've
read. You have to use it properly
and know what you are doing or better still, have RT extensions added to it.
I don't know what I am doing
(in this context) but I am trying to find out. All I know is that using
vanilla Windows in the way I am used to,
I'll get unpredictable response times which will occasionally be
catastophically unacceptable.
It's an interesting field that is for sure and I am rather surprised it is not
more popular. The industrial control field is massive but it appears that
using Windows for it is way beyond the means of many (most?) of the people
involved
in it. Heck, VB(Visual Basic) is the language of choice among my cohorts. Threads are what
you wear to work. Priorities are instructions to the shipping department.
And several thousand dollars for an RTOS that doesn't like Delphi and still
needs hardware to run on is hard to justify against an $800.00 plc that's
already as real-time as it gets (not to mention fast).
Anyway, I am still interested and I am reading just as fast as I can. :->
Jim Rowell
 

"Andrew Rybenkov" <XXXX@XXXXX.COM>writes
I'm concerned that there are many pitfalls and gotchas in simply setting
priority to real-time, etc. I get the impression that it is not easy or safe
hence the vast number of very expensive rtos systems and extensions
available for sale. Fun stuff though. Google, google, google......
 

"_real_ rtos cannot cost $100 a priori" ;))
something like Sony monitors that are overpriced just because they are Sony.
 

play Unreal Tourment in the same time?
 

"Jim Rowell" <XXXX@XXXXX.COM>writes
Windows solution if you can?
-d
 

Andrew Rybenkov writes:
that I used were not windows nor linux based, but I am quite sure you
would find reports in google about such things. If I recall well, the
multitasking is not "preemptive", but "collaborative" which is a bit
more difficult to program but way more predictible.
 

On Sun, 17 Oct 2004 04:11:22 +0400, "Andrew Rybenkov"
<XXXX@XXXXX.COM>writes:
(interrupts handled etc) will happen within set time limits, but that
means that overall, performance will be lower than for other OS'
without that guarantee.
At least thats my understanding on the topic.
- Asbjørn
 

If I recall well, the
your mouse is responsive always, except cases when some long task gets highest priority.
(and that was a major feature Bill Gates proudly announced about new OS, -
"you will never see frozen mouse again"), that is why you can terminate any process
via Task Manager.
Until then Windows was co-operative, so you had explicitly call ProcessMessages to let
other tasks to do something.
 

(interrupts handled etc) will happen within set time
limits. Rather, it is the programmer, through his
careful application planning that will guarantee such.
A "proper" RTOS tool can include planning, analysis,
simulation, compiler and linker measurement (like
checks for max recursion depth and required stack
size), and performance measurement tools that help
the programmer do this with confidence.
Some other features that I like to see in a "proper"
RTOS are (1) simplicity and intuitiveness so that the
programmers can learn it quickly have confidence in
their knowledge of it, (2) an interrupt priority
management scheme that does not put the lowest
priority H/W interrupt ahead of the highest priority
S/W task, (3) queuing schemes associated with the
task monitor (like the Ada rendezvous) for allowing
one task to provide another tasks with its next job,
and (4) the availability of a good OOP programming
language for use with it.
--JohnH