High frequency output on parallel port, how?

Hi,

I want to control stepper motors from XP, this requires me to toggle the
parallel port datalines with a frequency ofseveral KHz.
I have the givio driver which gives me IO privilege so that I
can write and read to the port in usermode.
Now I have the problem that the fastest timers in user mode is the
multimedia timer which works with 1 millisecond at most.
These guys http://www.artofcnc.ca claim that they can get 35 KHz output on
the parallel port of a 1 GHz machine running XP,
I know that they use a device driver, that's why I ask this question here.
How do I accomplish a timer that works with much higher frequencies, does
anybody of you kernel programmers know?

Bert

Robby
Tue Sep 21 16:29:08 CDT 2004

Hi,

I'm a controls engineer at the Canadian Light Source.

How fast do you need to go? One solution is to make a small circuit that
does some frequency multiplication. Probably not the sort of solution you
were looking for given the forum.

If you need it deterministic you could look at using RTEMS or RTLinux, we
use the former at our lab here for motor control. RTEMS is an open-source
real-time O/S developed by the DoD for missile guidance systems. It is
unix-like and has a VERY small footprint.

We have a mix of RTEMS, Linux and Windows inter-operating (grudgingly mind
you) on our controls network.

Cheers,
Rob




"Bert" <bert@nospam.home.nl> wrote in message
news:ciq37s$qh7$1@reader08.wxs.nl...
> Hi,
>
> I want to control stepper motors from XP, this requires me to toggle the
> parallel port datalines with a frequency ofseveral KHz.
> I have the givio driver which gives me IO privilege so that I
> can write and read to the port in usermode.
> Now I have the problem that the fastest timers in user mode is the
> multimedia timer which works with 1 millisecond at most.
> These guys http://www.artofcnc.ca claim that they can get 35 KHz output on
> the parallel port of a 1 GHz machine running XP,
> I know that they use a device driver, that's why I ask this question here.
> How do I accomplish a timer that works with much higher frequencies, does
> anybody of you kernel programmers know?
>
> Bert
>
>

Robby
Tue Sep 21 16:50:25 CDT 2004

You could also try the High Resolution Performance Counter.

Calling QueryPeformanceFrequency() on my PC gives 3579545 counts/second or
3.6 MHz (or 279 nS/tick). Knowing that, you can poll the performance
counter (QueryPerformanceCount()) adding up the time intervals until you get
the frequency you want. For example, to get 3.6 kHz poll until you have a
thousand ticks and then strobe your output. For a 50% duty cycle, you'd
need a bit more code and I'm not sure how much jitter is involved, but if
you want to do it all in user space, you could do some testing with the
perfomance counters.

Cheers,
Rob


"Bert" <bert@nospam.home.nl> wrote in message
news:ciq37s$qh7$1@reader08.wxs.nl...
> Hi,
>
> I want to control stepper motors from XP, this requires me to toggle the
> parallel port datalines with a frequency ofseveral KHz.
> I have the givio driver which gives me IO privilege so that I
> can write and read to the port in usermode.
> Now I have the problem that the fastest timers in user mode is the
> multimedia timer which works with 1 millisecond at most.
> These guys http://www.artofcnc.ca claim that they can get 35 KHz output on
> the parallel port of a 1 GHz machine running XP,
> I know that they use a device driver, that's why I ask this question here.
> How do I accomplish a timer that works with much higher frequencies, does
> anybody of you kernel programmers know?
>
> Bert
>
>

Maxim
Tue Sep 21 19:12:49 CDT 2004

> I want to control stepper motors from XP, this requires me to toggle the
> parallel port datalines with a frequency ofseveral KHz.

Forget this. The Windows OS cannot guarantee such a latency.

--
Maxim Shatskih, Windows DDK MVP
StorageCraft Corporation
maxim@storagecraft.com
http://www.storagecraft.com

Dave
Tue Sep 21 21:51:58 CDT 2004

Max is correct,
and Robby was close...

Jeez a $3 PIC micro that ya program in C and has a serial port to take you
own commands would solve this in a bout 5 minutes...

Dave

"Maxim S. Shatskih" <maxim@storagecraft.com> wrote in message
news:eKXafiDoEHA.1776@TK2MSFTNGP14.phx.gbl...
>> I want to control stepper motors from XP, this requires me to toggle the
>> parallel port datalines with a frequency ofseveral KHz.
>
> Forget this. The Windows OS cannot guarantee such a latency.
>
> --
> Maxim Shatskih, Windows DDK MVP
> StorageCraft Corporation
> maxim@storagecraft.com
> http://www.storagecraft.com
>
>

Neptune
Wed Sep 22 03:04:28 CDT 2004


"Robby Tanner" <rtanner@lightsource.ca> wrote in message
news:ulEH$TCoEHA.2904@TK2MSFTNGP15.phx.gbl...
> You could also try the High Resolution Performance Counter.
>
> Calling QueryPeformanceFrequency() on my PC gives 3579545 counts/second
or
> 3.6 MHz (or 279 nS/tick). Knowing that, you can poll the performance
> counter (QueryPerformanceCount()) adding up the time intervals until you
get
> the frequency you want. For example, to get 3.6 kHz poll until you have
a
> thousand ticks and then strobe your output. For a 50% duty cycle, you'd
> need a bit more code and I'm not sure how much jitter is involved, but
if
> you want to do it all in user space, you could do some testing with the
> perfomance counters.
>
> Cheers,
> Rob
>

I don't think QueryPeformanceFrequency() can control a parallel port
because its chipset has its frequency limitation.

He can generate a 35 KHz clock on an external oscillator circuit (using a
ripple counter, it's easy to do) then use NOR Gate or AND gate to
Open/Close the frequency, using the PC Parallel port to control the On/Off
state of that 35Khz frequency. If he wants to vary the output frequency he
can just toggle the parallel signal Up and Down with certain character
pattern. He will need to design a driving circuit any how to drive his
servomotor, so it makes sense for him to create an external clock along
with the servo driving circuit.

Cheers,

Neptune.

Bert
Wed Sep 22 10:59:08 CDT 2004

Sorry Guys,

You are all not close,
The guys in the link below make a CNC control
with stepper motors on the parallel port.
A fiend of mine has this system with 3 axes connected,
there is no external time what so ever.

The step output to the steppermotors really generate well above 10 khz
on hist 1.5 GHz machine. The steppers run fine, without lost pulses etc.
Everything on windows keeps running normally, no problem.
All I know is that a device driver is installed which probably does the
trick.

The question is how.
Some of the Guys proposed polling with QueryPerformanceCounter, but that
would
take all processor time and that is surely not the intention.

Lets wait for somone who really knows how to do it.

Regards,
Bert



"Bert" <bert@nospam.home.nl> schreef in bericht
news:ciq37s$qh7$1@reader08.wxs.nl...
> Hi,
>
> I want to control stepper motors from XP, this requires me to toggle the
> parallel port datalines with a frequency ofseveral KHz.
> I have the givio driver which gives me IO privilege so that I
> can write and read to the port in usermode.
> Now I have the problem that the fastest timers in user mode is the
> multimedia timer which works with 1 millisecond at most.
> These guys http://www.artofcnc.ca claim that they can get 35 KHz output on
> the parallel port of a 1 GHz machine running XP,
> I know that they use a device driver, that's why I ask this question here.
> How do I accomplish a timer that works with much higher frequencies, does
> anybody of you kernel programmers know?
>
> Bert
>
>

Robby
Wed Sep 22 11:57:10 CDT 2004

"Bert" <bert@nospam.home.nl> wrote in message
news:cis83c$kf0$1@reader08.wxs.nl...

> Lets wait for somone who really knows how to do it.

None of the potential solutions provided were "wrong". We have stepper
motors here that are controlling millions upon millions of dollars worth of
equipment with very fine tolerances. We have designed, prototyped and
fabricated our own custom controller boards because we couldn't find a
commercial one that met all our requirements. The technology has been
licensed to another company for production. I think I can safely say that
we "really know how to do it".

All we provided were possible solutions. We weren't pretending to know
exactly how this particular implementation is realized.

1) How exactly do you know the output frequency on the pins at the parallel
port? Where exactly on the link you provided did you get the 35 kHz value
from? Please send a link that that shows this claim and a description of
the product.

2) Is 35 kHz a requirement. If so why? I don't recall there being a
minimum frequency for running stepper motors; a maximum possibly but no
minimum. In fact, many times we'll send a single pulse to a motor at a
time. Your goal may not be so much a requirement as you think. More often,
the challenge is to provide a constant velocity, in which case WIndows XP is
out and probably so is any parallel-port only implementation.

Without thoroughly understanding what it is your trying to do, we can't help
you get there.

Rob

Dave
Wed Sep 22 17:39:08 CDT 2004

Not close huh.. Hmmm... Gee I guess the PIC I have taking commands from a PC
over a serial port doing PWM at 80KHZ driving a 500 Watt CO2 laser and
blowing holes in steel just isn't working... I better tell the customer
that it's broken.... :-)

Dave August

"Bert" <bert@nospam.home.nl> wrote in message
news:cis83c$kf0$1@reader08.wxs.nl...
> Sorry Guys,
>
> You are all not close,
> The guys in the link below make a CNC control
> with stepper motors on the parallel port.
> A fiend of mine has this system with 3 axes connected,
> there is no external time what so ever.
>
> The step output to the steppermotors really generate well above 10 khz
> on hist 1.5 GHz machine. The steppers run fine, without lost pulses etc.
> Everything on windows keeps running normally, no problem.
> All I know is that a device driver is installed which probably does the
> trick.
>
> The question is how.
> Some of the Guys proposed polling with QueryPerformanceCounter, but that
> would
> take all processor time and that is surely not the intention.
>
> Lets wait for somone who really knows how to do it.
>
> Regards,
> Bert
>
>
>
> "Bert" <bert@nospam.home.nl> schreef in bericht
> news:ciq37s$qh7$1@reader08.wxs.nl...
>> Hi,
>>
>> I want to control stepper motors from XP, this requires me to toggle the
>> parallel port datalines with a frequency ofseveral KHz.
>> I have the givio driver which gives me IO privilege so that I
>> can write and read to the port in usermode.
>> Now I have the problem that the fastest timers in user mode is the
>> multimedia timer which works with 1 millisecond at most.
>> These guys http://www.artofcnc.ca claim that they can get 35 KHz output
>> on
>> the parallel port of a 1 GHz machine running XP,
>> I know that they use a device driver, that's why I ask this question
>> here.
>> How do I accomplish a timer that works with much higher frequencies, does
>> anybody of you kernel programmers know?
>>
>> Bert
>>
>>
>
>

J
Thu Sep 23 02:42:31 CDT 2004

Bert <bert@nospam.home.nl> wrote:
> Sorry Guys,
>
> You are all not close,
> The guys in the link below make a CNC control
> with stepper motors on the parallel port.
> A fiend of mine has this system with 3 axes connected,
> there is no external time what so ever.
>
> The step output to the steppermotors really generate well above 10 khz
> on hist 1.5 GHz machine. The steppers run fine, without lost pulses
> etc. Everything on windows keeps running normally, no problem.
> All I know is that a device driver is installed which probably does
> the trick.

Why didn't you dig the site yourself ? It took me about 3 minutes to
find and RTFM at http://www.artofcnc.ca/Mach1.pdf:

[quote]
CNC under Windows platforms has long been thought to be difficult if not
impossible due to unpredictable interrupt and timing systems. Mach1 is
designed to avail itself of interrupt zero processing which allows it to
bypass Microsoft?s limitations in this respect. When interrupt zero
takeover of the 8259 PIC is deemed impossible due to different
motherboard hardware the program automatically attempts to take over
interrupt 255 which is the highest priority interrupt on local PIC
interrupting systems and use the Local APIC timer to instantiate a
real-time subsystem under Windows.
[snip]
The Mach1 driver is written as a form of virus. It is not a destructive
virus, but in order to create a real-time subsystem, it is necessary for
Mach1 to create code for timing in memory and then modify that code as
it checks your computer to re-write the code to match your system.
Windows does not allow this to occur unless a special security key is
generated to inform Windows that this type of self-modifying code is
allowed in your system. A reboot is necessary to instantiate this change
into your system. Without the reboot, when the code tries to self
modify, you will instantly get a Blue Screen of death and the security
key will be lost and the driver corrupted. Since the driver was loaded,
it will reload on the next boot which will signal a corrupt file to the
disk checking system and the process must be started again.
[end of quote]

This relates to Mach1, but this is probably applicable to other
products.
--
Jean-Paul Iribarren (aka JPI)
Anti-spam: ma véritable adresse électronique ne comporte pas de
chiffres;donc...
Anti-spam: my real e-mail address has no digits; consequently...

Robby
Thu Sep 23 17:36:13 CDT 2004

Further to what has been posted already, I got this from the developer (both
excerpts reprinted with permission):

"Mach2 takes over the scheduling. It uses the higest interrupt priority and
allows windows to see only its own time slices. The pulse stream is very
stable. I'm not using the high perfomrance timer or any Windows functions
for the timing , its all done in Ring0 and in full control of the system
hardware. Windows CAN take control, but generally only if a multimedia
application tries to use certain functions. Generally its not a problem.."

and when I asked about the performance hit:

"Performance varies form system to system. Personally, I answer mail and
such with no trouble, I know some who watch TV onscreen with no trouble. The
end effect is Windows thinks its running a slower processor, it sees no
other thing going on. A 1Ghz machine may , for example, perform like a
500Mhz machine when Mach2 is running as Mach2 may be steraling up to 50% of
its clock cycles. Depends highly on the computer though, alot of people
swear by using a clean install with only Mach2 running, my computer has
application from the last 7 years loaded and I run several while Mach2
runs."

Cheers,
Rob


"J.P. Iribarren" <famirib33@free.fr> wrote in message
news:cituje$m0n$1@s5.feed.news.oleane.net...
> Bert <bert@nospam.home.nl> wrote:
> > Sorry Guys,
> >
> > You are all not close,
> > The guys in the link below make a CNC control
> > with stepper motors on the parallel port.
> > A fiend of mine has this system with 3 axes connected,
> > there is no external time what so ever.
> >
> > The step output to the steppermotors really generate well above 10 khz
> > on hist 1.5 GHz machine. The steppers run fine, without lost pulses
> > etc. Everything on windows keeps running normally, no problem.
> > All I know is that a device driver is installed which probably does
> > the trick.
>
> Why didn't you dig the site yourself ? It took me about 3 minutes to
> find and RTFM at http://www.artofcnc.ca/Mach1.pdf:
>
> [quote]
> CNC under Windows platforms has long been thought to be difficult if not
> impossible due to unpredictable interrupt and timing systems. Mach1 is
> designed to avail itself of interrupt zero processing which allows it to
> bypass Microsoft?s limitations in this respect. When interrupt zero
> takeover of the 8259 PIC is deemed impossible due to different
> motherboard hardware the program automatically attempts to take over
> interrupt 255 which is the highest priority interrupt on local PIC
> interrupting systems and use the Local APIC timer to instantiate a
> real-time subsystem under Windows.
> [snip]
> The Mach1 driver is written as a form of virus. It is not a destructive
> virus, but in order to create a real-time subsystem, it is necessary for
> Mach1 to create code for timing in memory and then modify that code as
> it checks your computer to re-write the code to match your system.
> Windows does not allow this to occur unless a special security key is
> generated to inform Windows that this type of self-modifying code is
> allowed in your system. A reboot is necessary to instantiate this change
> into your system. Without the reboot, when the code tries to self
> modify, you will instantly get a Blue Screen of death and the security
> key will be lost and the driver corrupted. Since the driver was loaded,
> it will reload on the next boot which will signal a corrupt file to the
> disk checking system and the process must be started again.
> [end of quote]
>
> This relates to Mach1, but this is probably applicable to other
> products.
> --
> Jean-Paul Iribarren (aka JPI)
> Anti-spam: ma véritable adresse électronique ne comporte pas de
> chiffres;donc...
> Anti-spam: my real e-mail address has no digits; consequently...
>

Bert
Sun Sep 26 11:02:36 CDT 2004

Thanks Robby,

Is that very difficult to do?
I am not really a experienced devicedriver programmer, and could use some
good sample code that shows me how too get the highest priority timer
interrupt.

Bert


"Robby Tanner" <rtanner@lightsource.ca> schreef in bericht
news:uR4652boEHA.3840@TK2MSFTNGP09.phx.gbl...
> Further to what has been posted already, I got this from the developer
> (both
> excerpts reprinted with permission):
>
> "Mach2 takes over the scheduling. It uses the higest interrupt priority
> and
> allows windows to see only its own time slices. The pulse stream is very
> stable. I'm not using the high perfomrance timer or any Windows functions
> for the timing , its all done in Ring0 and in full control of the system
> hardware. Windows CAN take control, but generally only if a multimedia
> application tries to use certain functions. Generally its not a problem.."
>
> and when I asked about the performance hit:
>
> "Performance varies form system to system. Personally, I answer mail and
> such with no trouble, I know some who watch TV onscreen with no trouble.
> The
> end effect is Windows thinks its running a slower processor, it sees no
> other thing going on. A 1Ghz machine may , for example, perform like a
> 500Mhz machine when Mach2 is running as Mach2 may be steraling up to 50%
> of
> its clock cycles. Depends highly on the computer though, alot of people
> swear by using a clean install with only Mach2 running, my computer has
> application from the last 7 years loaded and I run several while Mach2
> runs."
>
> Cheers,
> Rob
>
>
> "J.P. Iribarren" <famirib33@free.fr> wrote in message
> news:cituje$m0n$1@s5.feed.news.oleane.net...
>> Bert <bert@nospam.home.nl> wrote:
>> > Sorry Guys,
>> >
>> > You are all not close,
>> > The guys in the link below make a CNC control
>> > with stepper motors on the parallel port.
>> > A fiend of mine has this system with 3 axes connected,
>> > there is no external time what so ever.
>> >
>> > The step output to the steppermotors really generate well above 10 khz
>> > on hist 1.5 GHz machine. The steppers run fine, without lost pulses
>> > etc. Everything on windows keeps running normally, no problem.
>> > All I know is that a device driver is installed which probably does
>> > the trick.
>>
>> Why didn't you dig the site yourself ? It took me about 3 minutes to
>> find and RTFM at http://www.artofcnc.ca/Mach1.pdf:
>>
>> [quote]
>> CNC under Windows platforms has long been thought to be difficult if not
>> impossible due to unpredictable interrupt and timing systems. Mach1 is
>> designed to avail itself of interrupt zero processing which allows it to
>> bypass Microsoft's limitations in this respect. When interrupt zero
>> takeover of the 8259 PIC is deemed impossible due to different
>> motherboard hardware the program automatically attempts to take over
>> interrupt 255 which is the highest priority interrupt on local PIC
>> interrupting systems and use the Local APIC timer to instantiate a
>> real-time subsystem under Windows.
>> [snip]
>> The Mach1 driver is written as a form of virus. It is not a destructive
>> virus, but in order to create a real-time subsystem, it is necessary for
>> Mach1 to create code for timing in memory and then modify that code as
>> it checks your computer to re-write the code to match your system.
>> Windows does not allow this to occur unless a special security key is
>> generated to inform Windows that this type of self-modifying code is
>> allowed in your system. A reboot is necessary to instantiate this change
>> into your system. Without the reboot, when the code tries to self
>> modify, you will instantly get a Blue Screen of death and the security
>> key will be lost and the driver corrupted. Since the driver was loaded,
>> it will reload on the next boot which will signal a corrupt file to the
>> disk checking system and the process must be started again.
>> [end of quote]
>>
>> This relates to Mach1, but this is probably applicable to other
>> products.
>> --
>> Jean-Paul Iribarren (aka JPI)
>> Anti-spam: ma véritable adresse électronique ne comporte pas de
>> chiffres;donc...
>> Anti-spam: my real e-mail address has no digits; consequently...
>>
>
>

Robby
Sun Sep 26 15:21:42 CDT 2004

I suspect it isn't trivial, although the author of the Mach2 driver claims
not to be a very experienced driver developer. He did mention that it took
a long time. If you need something of that nature, I would consider
purchasing Art's (the author of Mach2) or a similar production driver.

To the more experienced on the list:

out of academic curiousity, how would a driver take over the scheduling? I
read the explanation that JPI dug up but can't quite get my head around it.
Does the explanation make sense? Where would I read about the 8259 PIC?

Rob



"Bert" <bert@nospam.home.nl> wrote in message
news:cj6pf9$int$1@reader10.wxs.nl...
> Thanks Robby,
>
> Is that very difficult to do?
> I am not really a experienced devicedriver programmer, and could use some
> good sample code that shows me how too get the highest priority timer
> interrupt.
>
> Bert
>

Pavel
Mon Sep 27 16:03:17 CDT 2004

"Robby Tanner" <rtanner@lightsource.ca> wrote in message news:#IDjuZApEHA.3520@TK2MSFTNGP11.phx.gbl...
> out of academic curiousity, how would a driver take over the scheduling? I
> read the explanation that JPI dug up but can't quite get my head around it.
> Does the explanation make sense? Where would I read about the 8259 PIC?

Are you sure you want to read about 8259? Contemporary PCs don't contain this
antiquity any longer. Better read on APICs.

--PA

Robby
Mon Sep 27 18:18:00 CDT 2004

Fair enough, where do I read about APICs?

The approach used seems to grab INT 0 and effectively high jack the
scheduler. Are there any comments on the strategy?

Rob



"Pavel A." <pavel_a@geeklife.com> wrote in message
news:uJsfM5NpEHA.2504@TK2MSFTNGP10.phx.gbl...
> "Robby Tanner" <rtanner@lightsource.ca> wrote in message
news:#IDjuZApEHA.3520@TK2MSFTNGP11.phx.gbl...
> > out of academic curiousity, how would a driver take over the scheduling?
I
> > read the explanation that JPI dug up but can't quite get my head around
it.
> > Does the explanation make sense? Where would I read about the 8259 PIC?
>
> Are you sure you want to read about 8259? Contemporary PCs don't contain
this
> antiquity any longer. Better read on APICs.
>
> --PA
>
>
>

Robby
Tue Oct 05 16:22:58 CDT 2004

Where would I find details on using APICs, or a specification, etc?

Is it a chip, memory space, kernel object, etc? I'd like to read a
high-level explanation.

Cheers,
Rob


"Pavel A." <pavel_a@geeklife.com> wrote in message
news:uJsfM5NpEHA.2504@TK2MSFTNGP10.phx.gbl...
> "Robby Tanner" <rtanner@lightsource.ca> wrote in message
news:#IDjuZApEHA.3520@TK2MSFTNGP11.phx.gbl...
> > out of academic curiousity, how would a driver take over the scheduling?
I
> > read the explanation that JPI dug up but can't quite get my head around
it.
> > Does the explanation make sense? Where would I read about the 8259 PIC?
>
> Are you sure you want to read about 8259? Contemporary PCs don't contain
this
> antiquity any longer. Better read on APICs.
>
> --PA
>
>
>

Robby
Tue Oct 05 16:28:14 CDT 2004

Never mind, I think I found something at Intels website. APIC is just the
Advanced PIC, is that right?

Rob


"Robby Tanner" <rtanner@lightsource.ca> wrote in message
news:Owzp6FyqEHA.596@TK2MSFTNGP11.phx.gbl...
> Where would I find details on using APICs, or a specification, etc?
>
> Is it a chip, memory space, kernel object, etc? I'd like to read a
> high-level explanation.
>
> Cheers,
> Rob
>
>
> "Pavel A." <pavel_a@geeklife.com> wrote in message
> news:uJsfM5NpEHA.2504@TK2MSFTNGP10.phx.gbl...
> > "Robby Tanner" <rtanner@lightsource.ca> wrote in message
> news:#IDjuZApEHA.3520@TK2MSFTNGP11.phx.gbl...
> > > out of academic curiousity, how would a driver take over the
scheduling?
> I
> > > read the explanation that JPI dug up but can't quite get my head
around
> it.
> > > Does the explanation make sense? Where would I read about the 8259
PIC?
> >
> > Are you sure you want to read about 8259? Contemporary PCs don't contain
> this
> > antiquity any longer. Better read on APICs.
> >
> > --PA
> >
> >
> >
>
>

Arkady
Fri Oct 08 14:29:10 CDT 2004

Yes , it's allow much more IRQs ( 64 ) than 15 of PIC
Arkady

"Robby Tanner" <rtanner@lightsource.ca> wrote in message
news:elH52IyqEHA.3464@TK2MSFTNGP14.phx.gbl...
> Never mind, I think I found something at Intels website. APIC is just the
> Advanced PIC, is that right?
>
> Rob
>
>
> "Robby Tanner" <rtanner@lightsource.ca> wrote in message
> news:Owzp6FyqEHA.596@TK2MSFTNGP11.phx.gbl...
> > Where would I find details on using APICs, or a specification, etc?
> >
> > Is it a chip, memory space, kernel object, etc? I'd like to read a
> > high-level explanation.
> >
> > Cheers,
> > Rob
> >
> >
> > "Pavel A." <pavel_a@geeklife.com> wrote in message
> > news:uJsfM5NpEHA.2504@TK2MSFTNGP10.phx.gbl...
> > > "Robby Tanner" <rtanner@lightsource.ca> wrote in message
> > news:#IDjuZApEHA.3520@TK2MSFTNGP11.phx.gbl...
> > > > out of academic curiousity, how would a driver take over the
> scheduling?
> > I
> > > > read the explanation that JPI dug up but can't quite get my head
> around
> > it.
> > > > Does the explanation make sense? Where would I read about the 8259
> PIC?
> > >
> > > Are you sure you want to read about 8259? Contemporary PCs don't
contain
> > this
> > > antiquity any longer. Better read on APICs.
> > >
> > > --PA
> > >
> > >
> > >
> >
> >
>
>