Re: [Emc-users] Dedicated hardware for Real Time OS

[Ray Henry]

On Mon, 2008-11-24 at 07:30 -0800, Dave Engvall wrote:

> > IIRC Ray Henry is using some mini-itx boards with a C7 processsor  
> > and VIA's Unichrome(?) video. Apparently the on-board video
> > does not seriously interfere with the rt performance. It might be  
> > worth having Ray relate his experience.
> >
> > Dave
> > 

I've got two systems using MiniITX mobos here and duplicates of them at
the shop.  At the moment these are all running step-and-direction
drivers.  Latency specs for both are listed on the wiki.  I use the
onboard graphics and have very little problems with them.  In fact one
of the two rates right near best for those tested.

The combo display/mobo are really nice looking but damned pricy for any
of them I've checked out.  I built a cheap alternative using an
automotive 7" wide screen.

HTH

Rayh




-
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Dedicated hardware for Real Time OS

[Lester Caine]

Jack Coats wrote:
> I checked out the various ITX cards, and they all have one flaw, video 
> on board, and sometimes
> no parallel port, or way to put a 'bus' on them.  Video, ethernet, 
> parallel or other interface
> on-board makes for a non-optimum selection IMHO.  Small size is nice, 
> but being able to swap
> out a board with something off the shelf is important to non-hobby shops.
> 
> But the ITX boards are fascinating.  As you go from the larger to the 
> smaller, their price goes
> up quickly, from what I have seen.

CN13000 has the parallel port on an in board connector. Works fine 
driving Mach3 on XP and I'm not seeing any problems with Linux on them 
but have not yet sorted a new machine with the latest EMC2 on one. 
Previous tests with slower ITX boards worked fine with EMC2 but failed 
to handle Mach3/W2k. The key seems to be the C7 processor.

-- 
Lester Caine - G8HFL
-
Contact - http://lsces.co.uk/lsces/wiki/?page=contact
L.S.Caine Electronic Services - http://lsces.co.uk
EnquirySolve - http://enquirysolve.com/
Model Engineers Digital Workshop - http://medw.co.uk//
Firebird - http://www.firebirdsql.org/index.php

-
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Dedicated hardware for Real Time OS

[Dave Engvall]

On Nov 24, 2008, at 3:26 AM, Jack Coats wrote:

> I checked out the various ITX cards, and they all have one flaw, video
> on board, and sometimes
> no parallel port, or way to put a 'bus' on them.  Video, ethernet,
> parallel or other interface
> on-board makes for a non-optimum selection IMHO.  Small size is nice,
> but being able to swap
> out a board with something off the shelf is important to non-hobby  
> shops.
>
> But the ITX boards are fascinating.  As you go from the larger to the
> smaller, their price goes
> up quickly, from what I have seen.
>
> Erik Christiansen wrote:
>> On Sat, Nov 22, 2008 at 07:15:04PM -0800, Kirk Wallace wrote:
>>
>>> My two cents. I'd like to see what could be done with just a CPU,  
>>> RAM,
>>> Ethernet and GPIO, doing just the real time stuff. Maybe with four
>>> channels of signals for BLDC motors. One problem though is that  
>>> for a
>>> board like this, you can get a whole PC for less money. Too bad  
>>> cheap
>>> wifi routers don't have lots of GPIO. Making 200,000 EMC2 motion  
>>> boards
>>> _would_ bring the unit cost down.
>>>
>>
>> +1
>>
>> For some time now, I've been checking each LinuxDevices newsletter  
>> for
>> potentially suitable new mini-ITX cards. I've been looking for RS485
>> too, because that'd be great for intra-machine communication, if
>> satellite microcontollers are ever added, e.g. for a tool changer.  
>> (The
>> noise immunity of balanced line, and immunity to significant DC  
>> offset,
>> are both invaluable in the machine environment. It's also more  
>> real-time
>> than the buffered USB appears to be.)
>>
>> Fanless would be best, perhaps, so the enclosure could be sealed.
>>
>> Even if "just the real time stuff" is sufficiently modular to be  
>> run on
>> an SBC, it would be a lot of work to port e.g. eCos as underlying  
>> RTOS.
>> (I've only ever done that once, and am glad I was paid for it. ;-)
>>
>> Still, having a reference "EMC2 RT Platform" would eliminate parport
>> and other hardware compatibility worries. In time. But planning is  
>> the
>> first step.
>>
>> Erik
>>
>>
> IIRC Ray Henry is using some mini-itx boards with a C7 processsor  
> and VIA's Unichrome(?) video. Apparently the on-board video
> does not seriously interfere with the rt performance. It might be  
> worth having Ray relate his experience.
>
> Dave
> -- 
> ---
> This SF.Net email is sponsored by the Moblin Your Move Developer's  
> challenge
> Build the coolest Linux based applications with Moblin SDK & win  
> great prizes
> Grand prize is a trip for two to an Open Source event anywhere in  
> the world
> http://moblin-contest.org/redirect.php?banner_id=100&url=/
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users


-
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Dedicated hardware for Real Time OS

[Jack Coats]

I checked out the various ITX cards, and they all have one flaw, video 
on board, and sometimes
no parallel port, or way to put a 'bus' on them.  Video, ethernet, 
parallel or other interface
on-board makes for a non-optimum selection IMHO.  Small size is nice, 
but being able to swap
out a board with something off the shelf is important to non-hobby shops.

But the ITX boards are fascinating.  As you go from the larger to the 
smaller, their price goes
up quickly, from what I have seen.

Erik Christiansen wrote:
> On Sat, Nov 22, 2008 at 07:15:04PM -0800, Kirk Wallace wrote:
>   
>> My two cents. I'd like to see what could be done with just a CPU, RAM,
>> Ethernet and GPIO, doing just the real time stuff. Maybe with four
>> channels of signals for BLDC motors. One problem though is that for a
>> board like this, you can get a whole PC for less money. Too bad cheap
>> wifi routers don't have lots of GPIO. Making 200,000 EMC2 motion boards
>> _would_ bring the unit cost down.
>> 
>
> +1
>
> For some time now, I've been checking each LinuxDevices newsletter for
> potentially suitable new mini-ITX cards. I've been looking for RS485
> too, because that'd be great for intra-machine communication, if
> satellite microcontollers are ever added, e.g. for a tool changer. (The
> noise immunity of balanced line, and immunity to significant DC offset,
> are both invaluable in the machine environment. It's also more real-time
> than the buffered USB appears to be.)
>
> Fanless would be best, perhaps, so the enclosure could be sealed.
>
> Even if "just the real time stuff" is sufficiently modular to be run on
> an SBC, it would be a lot of work to port e.g. eCos as underlying RTOS.
> (I've only ever done that once, and am glad I was paid for it. ;-)
>
> Still, having a reference "EMC2 RT Platform" would eliminate parport
> and other hardware compatibility worries. In time. But planning is the
> first step.
>
> Erik
>
>   

-
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Dedicated hardware for Real Time OS

[Stephen Wille Padnos]

Erik Christiansen wrote:

>[snip]
>
>It's not necessarily the alternative that everyone would rush toward, I
>think. A small card with ethernet, parallel I/O, and preferably RS485,
>with floating point maths coprocessor, and a couple of PCI slots, is
>still the holy grail, I think. If it's economical, basic, and flexible,
>more of us can use it as the stone in stone soup.
>  
>
Economical and industrial PC don't belong in the same sentence.  Darn!

I looked up the price of a much much lesser device, an AMD Elan 
SC520-based unit with no graphics, no usable I/O, no discernible slots, 
and not much else (I don't know what the standard memory is, but for 
those unfamiliar with the SC520, it's more or less a glorified 486).  
That unit was priced at $1082 (it's here: 
).
  
The ones with LCDs are likely in the $3000+ range.  I looked at the 
specs for the panel PC style ones, and they all use the same chipsets as 
a consumer motherboard might.  The main difference is that they 
(probably) won't change the design every 3-6 months without telling 
you.  These systems are unlikely to have significantly better realtime 
performance than a random consumer PC.

I've looked into embedded PCs quite a lot, and I have a couple of them 
here for various projects.  I have one project that uses an embedded PC 
with Mesa FPGA card and some custom analog I/O cards.  It works for the 
task, but that's after extensive effort in reducing latency.  If 
graphics or network are used, latency spikes to 16000-ish.  This is on a 
machine that has <2000 under normal operation (and often under 200 ns).  
The chipset is spectacularly bad at PCI data transfer, with PCI 
transfers only about 2-3x the speed of parallel port I/O (with the 
exception that the address cycle is "instantaneous", and I get 32 bits 
per transfer instead of 8).

The point of all this text is to say that in my experience, industrial 
PCs are much more expensive, and rarely any better for RT, than consumer 
PCs.  They have features that may make them better for some 
applications, but RT performance is not one of them.  The few 
manufacturers that even mention RT are usually talking about 
milliseconds, not microseconds.  Those that are talking about 
microseconds are also providing a certified, often proprietary, OS.

Just my 2 cents.
- Steve


-
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Dedicated hardware for Real Time OS

[Erik Christiansen]

On Sun, Nov 23, 2008 at 12:23:22PM -0600, Mark Cason wrote:
> 
>   Or this:  http://linuxdevices.com/news/NS4282988768.html

That RTAI is ported is a big plus. Two PCI slots are essential. (One for
a graphics card, and one for parallel I/O, because there otherwise is
none!!?)

It has "VGA connector for resolutions up to 1280 x 1024", so on-board
graphics would need to be nobbled. Two CAN  opto-isolated interfaces would be
superfluous.

It's not necessarily the alternative that everyone would rush toward, I
think. A small card with ethernet, parallel I/O, and preferably RS485,
with floating point maths coprocessor, and a couple of PCI slots, is
still the holy grail, I think. If it's economical, basic, and flexible,
more of us can use it as the stone in stone soup.

Erik

-
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Dedicated hardware for Real Time OS

[Erik Christiansen]

On Sun, Nov 23, 2008 at 12:11:58PM -0600, Mark Cason wrote:
> 
>   How about something like this:
> 
>  http://www.linuxdevices.com/news/NS8140322559.html?kc=rss
> 
>   Still not perfect, but close.

Definitely a contender, subject to their "Via did not release pricing
for the device, ..." With no need for a separate LCD monitor, there's
room for a higher price. However, I had been looking more for a
(cheaper) small card to mount on the back of the machine (possibly in
one of some rugged ex-military enclosures I have). More important is not
to have all the I/O cabling running up to the display. (Short, direct,
armoured, shielded, and on the machine sounds good to me.) Then there's
only video and keyboard out there.

It would also be disappointing to spend a lot on a unit with fancy
on-board graphics, only to find that EMC2 FAQ_1 had bitten
unrelentingly.  ;-)

Erik

-
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Dedicated hardware for Real Time OS

[Gene Heskett]

On Sunday 23 November 2008, Mark Cason wrote:
>On Sun, 2008-11-23 at 16:35 +1100, Erik Christiansen wrote:
>> On Sat, Nov 22, 2008 at 07:15:04PM -0800, Kirk Wallace wrote:
>> > My two cents. I'd like to see what could be done with just a CPU, RAM,
>> > Ethernet and GPIO, doing just the real time stuff. Maybe with four
>> > channels of signals for BLDC motors. One problem though is that for a
>> > board like this, you can get a whole PC for less money. Too bad cheap
>> > wifi routers don't have lots of GPIO. Making 200,000 EMC2 motion boards
>> > _would_ bring the unit cost down.
>>
>> +1
>>
>> For some time now, I've been checking each LinuxDevices newsletter for
>> potentially suitable new mini-ITX cards. I've been looking for RS485
>> too, because that'd be great for intra-machine communication, if
>> satellite microcontollers are ever added, e.g. for a tool changer. (The
>> noise immunity of balanced line, and immunity to significant DC offset,
>> are both invaluable in the machine environment. It's also more real-time
>> than the buffered USB appears to be.)
>>
>> Fanless would be best, perhaps, so the enclosure could be sealed.
>>
>> Even if "just the real time stuff" is sufficiently modular to be run on
>> an SBC, it would be a lot of work to port e.g. eCos as underlying RTOS.
>> (I've only ever done that once, and am glad I was paid for it. ;-)
>>
>> Still, having a reference "EMC2 RT Platform" would eliminate parport
>> and other hardware compatibility worries. In time. But planning is the
>> first step.
>>
>> Erik
>
>  How about something like this:
>
> http://www.linuxdevices.com/news/NS8140322559.html?kc=rss
>
>  Still not perfect, but close.

With a parport, it would indeed be nice.  My relatively conventional box is on 
a high shelf, supposedly well above flying swarf even if I'm running an air 
sweep on the tooling.  That knocking sound you all know?  Exactly what you 
think it is. :)  However, the 10" screen is a bit small to sit in the present 
monitors location here.  However, at 8 pounds, I could hang it on a hinged 
bracket on the far side of the swarf shield.  Just below the box with the 
pmdx-106 spindle driver.

Interesting to be sure. Maybe they have a similar model with a parport?  But 
there is nothing like that on the site. :(

-- 
Cheers, Gene
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Root nameservers are out of sync

-
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Dedicated hardware for Real Time OS

[Mark Cason]

On Sun, 2008-11-23 at 12:11 -0600, Mark Cason wrote:
> On Sun, 2008-11-23 at 16:35 +1100, Erik Christiansen wrote:
> > On Sat, Nov 22, 2008 at 07:15:04PM -0800, Kirk Wallace wrote:
> > > My two cents. I'd like to see what could be done with just a CPU, RAM,
> > > Ethernet and GPIO, doing just the real time stuff. Maybe with four
> > > channels of signals for BLDC motors. One problem though is that for a
> > > board like this, you can get a whole PC for less money. Too bad cheap
> > > wifi routers don't have lots of GPIO. Making 200,000 EMC2 motion boards
> > > _would_ bring the unit cost down.
> > 
> > +1
> > 
> > For some time now, I've been checking each LinuxDevices newsletter for
> > potentially suitable new mini-ITX cards. I've been looking for RS485
> > too, because that'd be great for intra-machine communication, if
> > satellite microcontollers are ever added, e.g. for a tool changer. (The
> > noise immunity of balanced line, and immunity to significant DC offset,
> > are both invaluable in the machine environment. It's also more real-time
> > than the buffered USB appears to be.)
> > 
> > Fanless would be best, perhaps, so the enclosure could be sealed.
> > 
> > Even if "just the real time stuff" is sufficiently modular to be run on
> > an SBC, it would be a lot of work to port e.g. eCos as underlying RTOS.
> > (I've only ever done that once, and am glad I was paid for it. ;-)
> > 
> > Still, having a reference "EMC2 RT Platform" would eliminate parport
> > and other hardware compatibility worries. In time. But planning is the
> > first step.
> > 
> > Erik
> > 
> 
>   How about something like this:
> 
>  http://www.linuxdevices.com/news/NS8140322559.html?kc=rss
> 
>   Still not perfect, but close.
> 

  Or this:  http://linuxdevices.com/news/NS4282988768.html



-
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Dedicated hardware for Real Time OS

[Mark Cason]

On Sun, 2008-11-23 at 16:35 +1100, Erik Christiansen wrote:
> On Sat, Nov 22, 2008 at 07:15:04PM -0800, Kirk Wallace wrote:
> > My two cents. I'd like to see what could be done with just a CPU, RAM,
> > Ethernet and GPIO, doing just the real time stuff. Maybe with four
> > channels of signals for BLDC motors. One problem though is that for a
> > board like this, you can get a whole PC for less money. Too bad cheap
> > wifi routers don't have lots of GPIO. Making 200,000 EMC2 motion boards
> > _would_ bring the unit cost down.
> 
> +1
> 
> For some time now, I've been checking each LinuxDevices newsletter for
> potentially suitable new mini-ITX cards. I've been looking for RS485
> too, because that'd be great for intra-machine communication, if
> satellite microcontollers are ever added, e.g. for a tool changer. (The
> noise immunity of balanced line, and immunity to significant DC offset,
> are both invaluable in the machine environment. It's also more real-time
> than the buffered USB appears to be.)
> 
> Fanless would be best, perhaps, so the enclosure could be sealed.
> 
> Even if "just the real time stuff" is sufficiently modular to be run on
> an SBC, it would be a lot of work to port e.g. eCos as underlying RTOS.
> (I've only ever done that once, and am glad I was paid for it. ;-)
> 
> Still, having a reference "EMC2 RT Platform" would eliminate parport
> and other hardware compatibility worries. In time. But planning is the
> first step.
> 
> Erik
> 

  How about something like this:

 http://www.linuxdevices.com/news/NS8140322559.html?kc=rss

  Still not perfect, but close.


-
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Dedicated hardware for Real Time OS

[Greg Michalski]

Roberto - 

There seem to be a lot of replies to your topic but it somehow diverged into a 
developer conversation.

 I think what would help you most is the list that is on the wiki that people 
contribute latency test result numbers that are associated with their exact 
setup so that you can see what to expect if you find Brand X, Model Y 
motherboard with Z memory and A video, etc.  It's on the emc2 wiki - 
http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Latency-Test .  I think the 
information you're looking for is that list as you can see what to steer clear 
of and what gives really good test results.

That's the answer I'd give you - now my personal hardware situation - I am 
running an old P3 650MHz Gateway motherboard with onboard video and audio and I 
get a 13000 Max jitter on the fast thread which isnt wonderful, but it runs the 
mill (a Grizzly X3 using steppers on Gecko drives) just fine - I get 60ipm 
rapids with no issues running programs.  It's a bit slow load a LARGE program 
and is a little sluggish switching between workspaces (when I'm not running 
code I switch and hand edit  .ngc files in the 2nd workspace so EMC is still in 
the 1st workspace) but other than that it is fine.  I had originally planned to 
use a 2.4GHz Celeron in a VIA chipset MSI board but that got 18000 jitter.  My 
current desktop (which is soon to be decomissioned) gets a 7000 when I run the 
trial CD.  That has a 2.53GHz P4 3GB DDR, onboard 10/100/1000 LAN and using a 
Radeon 9600XT video card (forget the memory on the card) - once I finish all 
the other original parts of my mill (limit switches, pendant, etc.) I'll 
upgrade that desktop mobo into the mill since it's almost 2x as fast and 
doesn't show any sluggishness, and loads the same large .ngc files rapidly 
(nearly instant).  I admit I haven't entered my hardware info into the wiki 
list, which I should - the idea is to have as many people list what they've got 
and then people starting out have an idea of what they should look for.

HTH - sorry for the long winded response.

Greg
www.distinctperspectives.com

  - Original Message - 
  From: Roberto Caminiti 
  To: emc-users@lists.sourceforge.net 
  Sent: Friday, November 21, 2008 7:32 PM
  Subject: [Emc-users] Dedicated hardware for Real Time OS


  Hi everybody,
  I see that the newest hardware with frequency tuning and with many
  advanced features not always give the best performances in terms of
  latency, used with an RTOS.
  So now I will ask you...
  Are there on the market a special hardware designed to be used in
  Real Time environment? Where?
  So I think that this hardware is minimal, so I need only:
  - VGA
  - one or two PCI slot (for mesa cards)
  - Serial ports and/or parallel ports
  - One or two USB
  - LAN
  And so without advanced features like bluetooth, wi-fi, HDMI, audio, etc...

  Thank you!
  Roberto

  -
  This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
  Build the coolest Linux based applications with Moblin SDK & win great prizes
  Grand prize is a trip for two to an Open Source event anywhere in the world
  http://moblin-contest.org/redirect.php?banner_id=100&url=/
  ___
  Emc-users mailing list
  Emc-users@lists.sourceforge.net
  https://lists.sourceforge.net/lists/listinfo/emc-users



--



  No virus found in this incoming message.
  Checked by AVG - http://www.avg.com 
  Version: 8.0.175 / Virus Database: 270.9.9/1803 - Release Date: 11/21/2008 
9:37 AM
-
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Dedicated hardware for Real Time OS

[Erik Christiansen]

On Sat, Nov 22, 2008 at 07:15:04PM -0800, Kirk Wallace wrote:
> My two cents. I'd like to see what could be done with just a CPU, RAM,
> Ethernet and GPIO, doing just the real time stuff. Maybe with four
> channels of signals for BLDC motors. One problem though is that for a
> board like this, you can get a whole PC for less money. Too bad cheap
> wifi routers don't have lots of GPIO. Making 200,000 EMC2 motion boards
> _would_ bring the unit cost down.

+1

For some time now, I've been checking each LinuxDevices newsletter for
potentially suitable new mini-ITX cards. I've been looking for RS485
too, because that'd be great for intra-machine communication, if
satellite microcontollers are ever added, e.g. for a tool changer. (The
noise immunity of balanced line, and immunity to significant DC offset,
are both invaluable in the machine environment. It's also more real-time
than the buffered USB appears to be.)

Fanless would be best, perhaps, so the enclosure could be sealed.

Even if "just the real time stuff" is sufficiently modular to be run on
an SBC, it would be a lot of work to port e.g. eCos as underlying RTOS.
(I've only ever done that once, and am glad I was paid for it. ;-)

Still, having a reference "EMC2 RT Platform" would eliminate parport
and other hardware compatibility worries. In time. But planning is the
first step.

Erik

-- 
C hasn't changed much since the 1970s. And let's face it it's ugly.
Can't we do better? C++? (Sorry, never mind.)  - Rob Pike


-
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Dedicated hardware for Real Time OS

[Kirk Wallace]

My two cents. I'd like to see what could be done with just a CPU, RAM,
Ethernet and GPIO, doing just the real time stuff. Maybe with four
channels of signals for BLDC motors. One problem though is that for a
board like this, you can get a whole PC for less money. Too bad cheap
wifi routers don't have lots of GPIO. Making 200,000 EMC2 motion boards
_would_ bring the unit cost down.

Kirk
http://www.wallacecompany.com/machine_shop/



-
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Dedicated hardware for Real Time OS

[Jack Coats]

I am not suggesting 'throwing away' anything.  But as we look at going 
forward into a new
generation is something we might consider.

Going to a SBC that is really a mini-desktop mobo is probably generating 
more problems than
it is worth.  So doing something 'else' might be a more practical 
solution.  What the 'else' is, is up
to brighter minds than mine.

As you mentioned, some of the 'desktops' available today are 'encrusted' 
with more ports of various
flavors than are productive for EMC2 type of use.  I do believe we will 
have a number of 'hand-me-down'
system available for a number of years.  But before the source of 'good 
stuff' dries up, we need to
figure out what to do.

No action is required today.  I was just wanting folks to start thinking 
about it, for when the 'no legacy port'
days do eventually come.  Even it is more than a few years away.

Jon Elson wrote:
> Jack Coats wrote:
>   
>> For a different approach... Open source controller hardware with 
>> attached display.
>> 
>>
>> Another option might be to get one of the 'single board' Linux machines, 
>> use it to run
>> EMC, and do AXIS on a 'display computer'.  This might make it easier to 
>> do the 'port' and still
>> keep the 'real EMC2' on hardware without display (or sound).  It could 
>> use either USB or
>> Ethernet attachment.
>>
>> I could even see running Axis (or equivalent) on one of the little palm 
>> top computers, and EMC on
>> an 'attached controller box'.
>>
>> Yes, there is some hand waving and daydreaming in this, but one day we 
>> need to start thinking this way.
>>   
>> 
> My main question is "Why?"  Why throw away years of effort that has 
> produced a VERY powerful and flexible framework that WORKS?
> Is there a real time patch set for that "single board" Linux machine?  
> Without that, it is worthless for our purposes.  Why need TWO computers, 
> when one works fine?  Think of all the debugging and logging hassles 
> when you don't have hard drives on these computers, which would be the 
> case on many of the single board embedded systems.
>
> I certainly don't see the need for a move in this direction, yet.  
> Maybe, in the future, desktop PCs will be so horribly encrusted with 
> bluetooth and other wireless "crap", variable CPU speed, battery and 
> thermal management that they will be forever unable to do anything in 
> real time.  We may have to develop a hardware compatability list of our 
> own, testing specific models for RT performance, parallel port 
> compatability, etc.
> I can recommend a few models myself, I've been using off-lease Dell 
> Optiplex boxes with good results.
>
>
> Jon
>
> -
> This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
> Build the coolest Linux based applications with Moblin SDK & win great prizes
> Grand prize is a trip for two to an Open Source event anywhere in the world
> http://moblin-contest.org/redirect.php?banner_id=100&url=/
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>
>   

-
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Dedicated hardware for Real Time OS

[Michael Buesch]

On Saturday 22 November 2008 21:57:32 Roland Jollivet wrote:
> Hi
> 
> OK, sorry I posted that. I saw pretty pictures and did'nt read the specs.
> And I most often cringe when I see reply to my posts!!!
> 
> But, (here we go a gain;-) , I think there's a main philosophy difference
> with the guys who want an embedded controller. On the one hand, it's like
> the EMC crowd is primarily a Linux group, and the CNC part is integrated
> into their PC. As such, ever faster hardware is advocated (I concede, not
> needed) for EMC. The other group are building a CNC machine and need a
> controller, and may not be Linux users.
> As for two machines, well, most people already have one. They're using it
> right now,  to post.
> 
>  The CNC part is a seperate part; what they're building. I for one, battle
> with the advocation(latency)  "While the test is running, you should "abuse"
> the computer. Move windows around on the screen. Surf the web. Copy some
> large files..."
> I think to myself, but hey, this is my machine controller, I'd want to
> remove all that stuff for better speed. If I want to surf the net, I'll go
> sit in the office(or home)
> 
> A converse scenario is to imagine a machine operator standing at the console
> in front of his Haas, and he's busy googling global warming.

I don't see the problem. Often I'd _like_ to google for stuff while programming
on a CNC machine. With EMC I _can_ do that. I see this as a great advantage.

-- 
Greetings Michael.

-
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Dedicated hardware for Real Time OS

[Roland Jollivet]

Hi

OK, sorry I posted that. I saw pretty pictures and did'nt read the specs.
And I most often cringe when I see reply to my posts!!!

But, (here we go a gain;-) , I think there's a main philosophy difference
with the guys who want an embedded controller. On the one hand, it's like
the EMC crowd is primarily a Linux group, and the CNC part is integrated
into their PC. As such, ever faster hardware is advocated (I concede, not
needed) for EMC. The other group are building a CNC machine and need a
controller, and may not be Linux users.
As for two machines, well, most people already have one. They're using it
right now,  to post.

 The CNC part is a seperate part; what they're building. I for one, battle
with the advocation(latency)  "While the test is running, you should "abuse"
the computer. Move windows around on the screen. Surf the web. Copy some
large files..."
I think to myself, but hey, this is my machine controller, I'd want to
remove all that stuff for better speed. If I want to surf the net, I'll go
sit in the office(or home)

A converse scenario is to imagine a machine operator standing at the console
in front of his Haas, and he's busy googling global warming.

Anyway, as expected with open source project, people pull in all different
directions.

Regards
Roland Jollivet



2008/11/22 Jon Elson <[EMAIL PROTECTED]>

> Roland Jollivet wrote:
> > Hi
> > It would be nice to have on something like this;
> >
> > http://cubloc.com/product/05_02.php
> >
> Especially, no parallel port.  What about a disk drive?  SD card?
> > (with linux of course)
> >
> That probably exists.
> > or;
> >
> > http://www.linuxdevices.com/news/NS2304885763.html
> >
> No hard drive, no parallel port (although they mention GPIO but no idea
> how many signals.)
> No floating point, either, and EMC relies HEAVILY on floating point
> arithmetic.
> On both of these systems, the LCD has pretty low resolution, although
> Axis could probably be tweaked to work pretty well on it.
>
> Jon
>
> -
> This SF.Net email is sponsored by the Moblin Your Move Developer's
> challenge
> Build the coolest Linux based applications with Moblin SDK & win great
> prizes
> Grand prize is a trip for two to an Open Source event anywhere in the world
> http://moblin-contest.org/redirect.php?banner_id=100&url=/
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>
-
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Dedicated hardware for Real Time OS

[Rafael Skodlar]

Jon Elson wrote:
> Jack Coats wrote:
>> For a different approach... Open source controller hardware with 
>> attached display.
>> 
>>
>> Another option might be to get one of the 'single board' Linux machines, 
>> use it to run
>> EMC, and do AXIS on a 'display computer'.  This might make it easier to 
>> do the 'port' and still
>> keep the 'real EMC2' on hardware without display (or sound).  It could 
>> use either USB or
>> Ethernet attachment.
>>
>> I could even see running Axis (or equivalent) on one of the little palm 
>> top computers, and EMC on
>> an 'attached controller box'.
>>
>> Yes, there is some hand waving and daydreaming in this, but one day we 
>> need to start thinking this way.
>>   
> My main question is "Why?"  Why throw away years of effort that has 
> produced a VERY powerful and flexible framework that WORKS?

Exactly. Sending commands to CNC components (motors, sensors, etc.)
through USB attached micro controllers is too hard to support with
limited resources not to mention basic timing issues.

> Is there a real time patch set for that "single board" Linux machine?  
> Without that, it is worthless for our purposes.  Why need TWO computers, 
> when one works fine?  Think of all the debugging and logging hassles 
> when you don't have hard drives on these computers, which would be the 
> case on many of the single board embedded systems.
> 

I believe that "embedded CNC computer" is a viable option. It all
depends on what we take as embedded. With numerous mini itx boards on
the market today it should be possible to find one that's good enough
for controlling some CNC machines. CAD design etc. can all be done
elsewhere with files transported over the physical or wireless network,
or flash memory (USB, SD) to embedded CNC system.

It's better not to mess with the CNC system when CAD or some other
program needs an upgrade for example. I would not want to have somebody
browse the websites while CNC machine was running on the same system so
a dedicated computer makes sense.

> I certainly don't see the need for a move in this direction, yet.  

... yet. And that's the key. Being involved in USB memory devices at
work I can say with high confidence that years of classic hard drive for
basic computer storage are numbered. Let's face it, it's a mechanical
device not very suitable for a shop floor where temperatures and
vibrations reach over recommended ones. SSD (Solid State Drives) are
going to take over due to lower power, reliability, weight, etc., in few
years.

Like it or not, USB is going to play a major role in all it's current
and future forms in computing environment. Even high end servers are now
coming with the OS installed on SD cards for example. I see no reason
for EMC not to bootup from SD card and mostly run in memory with only
file or possibly video exchange needed over the network.

> Maybe, in the future, desktop PCs will be so horribly encrusted with 
> bluetooth and other wireless "crap", variable CPU speed, battery and 
> thermal management that they will be forever unable to do anything in 

While that might become a reality, the future for RT boards looks very
promising. Open source is going places including hardware. Take
http://beagleboard.org for example. I've seen a demo of that board
recently with focus on robotics.

>From what I learned, this is very powerful board with a lot of
possibilities. It seems that Ubuntu is going to be well presented on
that platform which makes me think that it will be suitable for CNC
eventually.

I'm not advocating this board as "the solution" for CNC as I'm not
familiar with it enough, just want to point out that new options are
opening as the old ones are fading away.

Who needs silly parallel port when real IO ports are desired? Don't we
have enough problems with PP in different modes, BIOS setup etc.?
Spending much time on PP port for serious RT is not worth it IMO when
low cost IO boards are available.

> real time.  We may have to develop a hardware compatability list of our 
> own, testing specific models for RT performance, parallel port 
> compatability, etc.
> I can recommend a few models myself, I've been using off-lease Dell 
> Optiplex boxes with good results.
> 
> 
> Jon
> 

--
Rafael

-
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Dedicated hardware for Real Time OS

[Jon Elson]

Roland Jollivet wrote:
> Hi
> It would be nice to have on something like this;
>
> http://cubloc.com/product/05_02.php
>   
Especially, no parallel port.  What about a disk drive?  SD card?
> (with linux of course)
>   
That probably exists.
> or;
>
> http://www.linuxdevices.com/news/NS2304885763.html
>   
No hard drive, no parallel port (although they mention GPIO but no idea 
how many signals.)
No floating point, either, and EMC relies HEAVILY on floating point 
arithmetic.
On both of these systems, the LCD has pretty low resolution, although 
Axis could probably be tweaked to work pretty well on it.

Jon

-
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Dedicated hardware for Real Time OS

[Jon Elson]

Jack Coats wrote:
> For a different approach... Open source controller hardware with 
> attached display.
> 
>
> Another option might be to get one of the 'single board' Linux machines, 
> use it to run
> EMC, and do AXIS on a 'display computer'.  This might make it easier to 
> do the 'port' and still
> keep the 'real EMC2' on hardware without display (or sound).  It could 
> use either USB or
> Ethernet attachment.
>
> I could even see running Axis (or equivalent) on one of the little palm 
> top computers, and EMC on
> an 'attached controller box'.
>
> Yes, there is some hand waving and daydreaming in this, but one day we 
> need to start thinking this way.
>   
My main question is "Why?"  Why throw away years of effort that has 
produced a VERY powerful and flexible framework that WORKS?
Is there a real time patch set for that "single board" Linux machine?  
Without that, it is worthless for our purposes.  Why need TWO computers, 
when one works fine?  Think of all the debugging and logging hassles 
when you don't have hard drives on these computers, which would be the 
case on many of the single board embedded systems.

I certainly don't see the need for a move in this direction, yet.  
Maybe, in the future, desktop PCs will be so horribly encrusted with 
bluetooth and other wireless "crap", variable CPU speed, battery and 
thermal management that they will be forever unable to do anything in 
real time.  We may have to develop a hardware compatability list of our 
own, testing specific models for RT performance, parallel port 
compatability, etc.
I can recommend a few models myself, I've been using off-lease Dell 
Optiplex boxes with good results.


Jon

-
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Dedicated hardware for Real Time OS

[Roland Jollivet]

Hi
It would be nice to have on something like this;

http://cubloc.com/product/05_02.php

(with linux of course)

or;

http://www.linuxdevices.com/news/NS2304885763.html

So the whole thing would hanng together like this;

http://controlafzar.com/products_en/controller_page.htm



Regards
Roland Jollivet


2008/11/22 Jack Coats <[EMAIL PROTECTED]>

> For a different approach... Open source controller hardware with
> attached display.
>
> Getting a stable real time environment are why ShopBotTools and
> FlashCutCNC both use
> USB attached controllers.  This is where the sensors and motors are all
> attached.
> Control happens in the USB attached computer, typically Windows running
> some
> client software that gives a display similar to Axis.
>
> Even the open source robotic LeafProject.org does something similar,
> using a PIC
> or similar board to control wheels and sensors, and the 'main computer'
> is a Windows
> laptop, in this case running a LISP interpreter.
>
> The CarveWright.com machine does something similar but has a dedicated
> controller
> and a SD card interface for loading programs.
>
> We could consider a small 'off board controller'.  Hopefully opensource
> like the
> Seguino control computer (A AMT micro processor that has a C like
> development
> kit available for free).  This particular controller has quite a few
> both digital and analog
> controls.  This might allow running EMC2 in a non-real-time
> environment.  I would still
> like to keep it unix/Linux compatible even if someone wants to do a Mac
> or Windows
> compatible port.
>
> Historically I remember even IBM made boxes with high-speed (for the
> day) floating
> point array processors where 'array programs and data' were loaded into
> these
> I/O attached units then they processed independently, and send the data
> back when done,
> and even 'attached CPUs' that had none of their own I/O but added raw
> processor
> cycles.  Even the venerable Intel 286 and 386 chips had floating point
> co-processors that
> were basically good calcularort chips attached.
>
> I am not suggesting a g-code interpreter, but that might be OK on a
> larger unit, but possibly
> a very simplified pseudo-g-code that could be used to implement a full
> g-code in conjunction
> with the control/display computer.
>
> Another option might be to get one of the 'single board' Linux machines,
> use it to run
> EMC, and do AXIS on a 'display computer'.  This might make it easier to
> do the 'port' and still
> keep the 'real EMC2' on hardware without display (or sound).  It could
> use either USB or
> Ethernet attachment.
>
> I could even see running Axis (or equivalent) on one of the little palm
> top computers, and EMC on
> an 'attached controller box'.
>
> Yes, there is some hand waving and daydreaming in this, but one day we
> need to start thinking this way.
>
>
> Gene Heskett wrote:
> > On Friday 21 November 2008, Roberto Caminiti wrote:
> >
> >> Hi everybody,
> >> I see that the newest hardware with frequency tuning and with many
> >> advanced features not always give the best performances in terms of
> >> latency, used with an RTOS.
> >> So now I will ask you...
> >> Are there on the market a special hardware designed to be used in
> >> Real Time environment? Where?
> >> So I think that this hardware is minimal, so I need only:
> >> - VGA
> >> - one or two PCI slot (for mesa cards)
> >> - Serial ports and/or parallel ports
> >> - One or two USB
> >> - LAN
> >> And so without advanced features like bluetooth, wi-fi, HDMI, audio,
> etc...
> >>
> >
> > The observations I can make are 1. onboard video is generally a bad thing
> for
> > realtime, particularly if it shares the mainboard memory.  It doesn't
> want to
> > give access back to the cpu unless threatened.
> >
> > And 2. having a storage device, like a usb key or similar, plugged into a
> usb
> > port, also mungs the realtime stuff because of the device scans being
> > performed to see if its still there at frequent intervals.
> >
> > There may be others, one of which could be a mainboard std voltage of 3.3
> > volts rather than 5.  The 3.3 volt parport setups seem to be pretty picky
> > about who they talk to.
> >
> >
> >> Thank you!
> >> Roberto
> >>
> >>
> -
> >> This SF.Net email is sponsored by the Moblin Your Move Developer's
> challenge
> >> Build the coolest Linux based applications with Moblin SDK & win great
> >> prizes Grand prize is a trip for two to an Open Source event anywhere in
> >> the world http://moblin-contest.org/redirect.php?banner_id=100&url=/
> >> ___
> >> Emc-users mailing list
> >> Emc-users@lists.sourceforge.net
> >> https://lists.sourceforge.net/lists/listinfo/emc-users
> >>
> >
> >
> >
> >
>
> -
> This SF.Net email is sponsored by the Moblin Your Move Developer's
> challenge
> Build the

Re: [Emc-users] Dedicated hardware for Real Time OS

[Jack Coats]

For a different approach... Open source controller hardware with 
attached display.

Getting a stable real time environment are why ShopBotTools and 
FlashCutCNC both use
USB attached controllers.  This is where the sensors and motors are all 
attached.
Control happens in the USB attached computer, typically Windows running some
client software that gives a display similar to Axis.

Even the open source robotic LeafProject.org does something similar, 
using a PIC
or similar board to control wheels and sensors, and the 'main computer' 
is a Windows
laptop, in this case running a LISP interpreter.

The CarveWright.com machine does something similar but has a dedicated 
controller
and a SD card interface for loading programs.

We could consider a small 'off board controller'.  Hopefully opensource 
like the
Seguino control computer (A AMT micro processor that has a C like 
development
kit available for free).  This particular controller has quite a few 
both digital and analog
controls.  This might allow running EMC2 in a non-real-time 
environment.  I would still
like to keep it unix/Linux compatible even if someone wants to do a Mac 
or Windows
compatible port.

Historically I remember even IBM made boxes with high-speed (for the 
day) floating
point array processors where 'array programs and data' were loaded into 
these
I/O attached units then they processed independently, and send the data 
back when done,
and even 'attached CPUs' that had none of their own I/O but added raw 
processor
cycles.  Even the venerable Intel 286 and 386 chips had floating point 
co-processors that
were basically good calcularort chips attached.

I am not suggesting a g-code interpreter, but that might be OK on a 
larger unit, but possibly
a very simplified pseudo-g-code that could be used to implement a full 
g-code in conjunction
with the control/display computer.

Another option might be to get one of the 'single board' Linux machines, 
use it to run
EMC, and do AXIS on a 'display computer'.  This might make it easier to 
do the 'port' and still
keep the 'real EMC2' on hardware without display (or sound).  It could 
use either USB or
Ethernet attachment.

I could even see running Axis (or equivalent) on one of the little palm 
top computers, and EMC on
an 'attached controller box'.

Yes, there is some hand waving and daydreaming in this, but one day we 
need to start thinking this way.


Gene Heskett wrote:
> On Friday 21 November 2008, Roberto Caminiti wrote:
>   
>> Hi everybody,
>> I see that the newest hardware with frequency tuning and with many
>> advanced features not always give the best performances in terms of
>> latency, used with an RTOS.
>> So now I will ask you...
>> Are there on the market a special hardware designed to be used in
>> Real Time environment? Where?
>> So I think that this hardware is minimal, so I need only:
>> - VGA
>> - one or two PCI slot (for mesa cards)
>> - Serial ports and/or parallel ports
>> - One or two USB
>> - LAN
>> And so without advanced features like bluetooth, wi-fi, HDMI, audio, etc...
>> 
>
> The observations I can make are 1. onboard video is generally a bad thing for 
> realtime, particularly if it shares the mainboard memory.  It doesn't want to 
> give access back to the cpu unless threatened. 
>
> And 2. having a storage device, like a usb key or similar, plugged into a usb 
> port, also mungs the realtime stuff because of the device scans being 
> performed to see if its still there at frequent intervals.
>
> There may be others, one of which could be a mainboard std voltage of 3.3 
> volts rather than 5.  The 3.3 volt parport setups seem to be pretty picky 
> about who they talk to.
>
>   
>> Thank you!
>> Roberto
>>
>> -
>> This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
>> Build the coolest Linux based applications with Moblin SDK & win great
>> prizes Grand prize is a trip for two to an Open Source event anywhere in
>> the world http://moblin-contest.org/redirect.php?banner_id=100&url=/
>> ___
>> Emc-users mailing list
>> Emc-users@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/emc-users
>> 
>
>
>
>   

-
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Dedicated hardware for Real Time OS

[Gene Heskett]

On Friday 21 November 2008, Roberto Caminiti wrote:
>Hi everybody,
>I see that the newest hardware with frequency tuning and with many
>advanced features not always give the best performances in terms of
>latency, used with an RTOS.
>So now I will ask you...
>Are there on the market a special hardware designed to be used in
>Real Time environment? Where?
>So I think that this hardware is minimal, so I need only:
>- VGA
>- one or two PCI slot (for mesa cards)
>- Serial ports and/or parallel ports
>- One or two USB
>- LAN
>And so without advanced features like bluetooth, wi-fi, HDMI, audio, etc...

The observations I can make are 1. onboard video is generally a bad thing for 
realtime, particularly if it shares the mainboard memory.  It doesn't want to 
give access back to the cpu unless threatened. 

And 2. having a storage device, like a usb key or similar, plugged into a usb 
port, also mungs the realtime stuff because of the device scans being 
performed to see if its still there at frequent intervals.

There may be others, one of which could be a mainboard std voltage of 3.3 
volts rather than 5.  The 3.3 volt parport setups seem to be pretty picky 
about who they talk to.

>Thank you!
>Roberto
>
>-
>This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
>Build the coolest Linux based applications with Moblin SDK & win great
> prizes Grand prize is a trip for two to an Open Source event anywhere in
> the world http://moblin-contest.org/redirect.php?banner_id=100&url=/
>___
>Emc-users mailing list
>Emc-users@lists.sourceforge.net
>https://lists.sourceforge.net/lists/listinfo/emc-users



-- 
Cheers, Gene
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Windows: Microsoft's tax on computer illiterates.

-
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

[Emc-users] Dedicated hardware for Real Time OS

[Roberto Caminiti]

Hi everybody,
I see that the newest hardware with frequency tuning and with many
advanced features not always give the best performances in terms of
latency, used with an RTOS.
So now I will ask you...
Are there on the market a special hardware designed to be used in
Real Time environment? Where?
So I think that this hardware is minimal, so I need only:
- VGA
- one or two PCI slot (for mesa cards)
- Serial ports and/or parallel ports
- One or two USB
- LAN
And so without advanced features like bluetooth, wi-fi, HDMI, audio, etc...

Thank you!
Roberto

-
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users