Re: real time linux

"John R. Carroll" <jcarroll@xxxxxxxxxxxxxxxxxxxxx> wrote in
news:Skptf.42338$q%.37320@xxxxxxxxxxxxxxxxxxxxxxxxxx:

> D Murphy wrote:
>> "John R. Carroll" <jcarroll@xxxxxxxxxxxxxxxxxxxxx> wrote in
>> news:b7ntf.48852$6e1.25806@xxxxxxxxxxxxxxxxxxxxxxxxxx:
>>
>> <snip> X 2
>>
>>
>> I still wonder about the patent. I looked it up and it seems to me at
>> least that these new controls would have to use technology which is
>> covered by the patent. Here is a link: <http://tinyurl.com/7bs3z> You
>> need a tiff plug-in for your browser to view the drawings.
>>
>> It could be that since all of the controls that are capable of
>> running compiled programs are European, Tornos has licenced the
>> technology. But even that seems unlikely. If there is some
>> fundemental difference in the control technology I'd be interested in
>> finding out what it is. I mean outside of the STEP standard and the
>> ability to edit and compile on the control. (An obvious advantage
>> IMO)
>
> What is described and embodied in this patent follows the classic
> scalar computing architectural model.
> What is currently in development is not scalar at all. It's vector.
> I'm not being a smart ass here Dan - do you understand the difference
> in these models?
> In lay, and not entirely accurate terms, the stuff I have seen doesn't
> use a "table" structure the way Tornos-Bechler indicates at all.
> Let me put it like this.
> A computer would tell you to drive a nail by
> A) Picking up the hammer
> B) Pick up the Nail
> C) Place the Nail against the wood.
> D) Lift the hammer
>
> And so on. This is a scalar sequential microprocessing model. It's the
> way your Tsugami machines work now. It may not look that way because
> things happen quickly and you can operate several channels at the same
> time but it none the less works just like this. The instructions are
> not really synchronous. They may be simultaneous but they are
> synchronized to an interval length.
>
> A vector architecture would issue all of the instructions at once, to
> be vectored to the appropriate address space and you'd be picking up
> the nail and hammer at the same time and then placing the nail and
> simultaneously moving the hammer and so forth. These instructions are
> evaluated asynchronously and set to complete at a specific point in
> space AND time independent of the actual path. That path is described
> using the kinematics model in the machine and other part and machine
> related parameters and at that point they are path dependent. You
> should be seeing something like this in the new FANUC's you are
> beginning to work on.
>
> It's the difference between set piece linear computing and artificial
> intelligence and with a vector machine you never either know or care
> what part of the calculation is being done where. Optimizing the work
> flow can be a bitcheroonie but that's just a matter of watch dogging
> the process in real time. Another plus is that the vectors hold more
> data that the scalars so you also get the inherant advantage of a
> reduced instruction set environment. The vectors hold most of the
> processing instructions. There is no table in other words. It works
> like your brain does.
>
> Anyway, farter smuckers than me could give you the real low down.
>
>

I've been doing a little more reading on the net and have come to the
conclusion that we are talking about apples and oranges in a certain sense.

What I'm familiar with is the technology where the program data is used to
create a compiled set of data. For the purpose of eliminating the
calculations while the machine cycle is being run. From what I've been able
to figure out after reading a bunch of pdfs on Siemen's web site is that
their controls allow "technology specific compile cycles", in other words
the architecture is wide open and a buider can write their own compilers
and add them to the basic software in the control.

This sounds consistant with the way I understand the Star works. Once
you've doped out your program you then run a compiling cycle, then run from
the compiled data. I've been told it will take about 15-20% off of the
cycle time.

So in order to get the information I'm interested in, namely what gets
compiled, I'd likely have to talk to the guy who developed the software at
the machine tool builder or get hold of some of their technical docs.


--

Dan

.

Relevant Pages

  • Re: How Can I disable autosaving when quitting Access?
    ... The compile option is disabled, I thought you could only compile if you're ... Doug Steele, Microsoft Access MVP ... that some of the controls are unboud such as the date pickers is because ... There are a few other anomalies of the sort (iCheck ...
    (microsoft.public.access.formscoding)
  • ASP.NET Compiler Behavior Affected by the Page Content?
    ... I've been struggling with some issues relating to referencing child controls ... and working and in that code, he references child controls within a FormView ... This code will not compile because MyComment is not a valid property. ...
    (microsoft.public.dotnet.framework.aspnet)
  • RE: Compile errors after converting to a Web Application Project from
    ... there are more restrictions on code sharing (web site compile ... partial classes (basically declaring server controls as protected global). ... i have never seen an advantage to switch to web applications, ... found at "Converting a Web Site Project to a Web Application Project". ...
    (microsoft.public.dotnet.framework.aspnet)
  • Re: CREATE
    ... possible code field address. ... defining word's definition. ... SEP instructions. ... The only difference being that, at compile time, VARIABLE ...
    (comp.lang.forth)
  • Re: Could an Intel 486 wake up with FC3
    ... > On the sidenote, since i586 is minimal supported platform, I don't know ... There were very few instructions added to the instruction set between ... relevant), system rebooted between runs, just different compile flags. ... James Wilkinson ...
    (Fedora)