foxchip wrote:
> On Aug 22, 10:00 am, John Passaniti
JapanIsShinto.com> wrote:>> programming environment that to me is euphemistically called Forth, but
>
> 'Called Forth' by the person who invented and named the Forth
> programming
> language. I do realize that there are people who actually claim that
> they
> know more about what is/isn't Forth than the inventor or who even say
> that the inventor should not use the term now that they own it. ;-)
> I also know who those people are in c.l.f. ;-)
As is typical with most of your arguments, you're making up claims. I
make no claim that I know Forth better than Charles Moore. I make no
statement that Charles Moore shouldn't use the term Forth however he
sees fit. If you want to create ridiculous straw men to fight against,
you go right ahead. It certainly is easier than defending real positions.
Things made a lot more sense when Charles Moore prefixed his work to
differentiate it from prior work. For example, saying "MachineForth"
meant something crisp and could be clearly discussed. Saying
"ColorForth" was unambiguous. Now it seems that with dropping the
prefix, "Forth" could mean anything spanning the history of Forth, from
Monsanto to whatever he's feeling like today.
>> The Tilera chips offer a C-based programming environment and the ability
>> to run Linux. Neither C or Linux means it is "better," but it does mean
>> that the learning curve to start using their chips is going to be lower.
>
> Indeed that is true assuming you assume C programmers, ie. programmers
> thinking
> in and using C will have lower learning curve for C than something
> else. Duh.
That's not my statement. When I looked over the available documentation
for the Tilera chips, one thing that struck me is that like the
IntellaSys chips, some aspects of the underlying hardware are exposed to
the programmer. But unlike the IntellaSys chips, it looks like they put
a lot of effort in abstracting that from the programmer-- at least if
the programmer is using C.
That is, if the programmer wants to just get some code up and running,
they can bring up the system with familiar tools and get something to
work. They *then* can start to dive in, look at partitioning their code
across the processing nodes, and using low-level aspects (like
communication delays) to their advantage.
The difference in the IntellaSys chips-- at least with what's documented
on the web site-- is that the programmer must start with a deep
awareness of all the quirks and design decisions of the architecture.
And again, this doesn't mean the Tilera chips are better. It does mean
however that a programmer starting with one is more likely to be able to
get the system up and running sooner. It's not an issue of using C or
Linux. It's an issue that they put the effort into creating tools that
allow the programmer to defer getting into the low-level details.
> If you assume Forth programmers the learning curve is lower with
> Forth.
> I would assume a mix of some Forth and some C programmers in a
> newgroup
> even if the name is 'comp.lang.forth.'
I guess I'll have to take your word for it. Forth programmers certainly
won't have a problem with the IntellaSys chips use of stacks. But when
they start getting into the ideas that aren't common with most Forth
programmers-- such as instruction scheduling and dealing with limits of
addressability based on which slot an instruction falls in, then those
are skills that I think are going to be challenging regardless of their
exposure to Forth.
>> A cynical interpretation is that they don't have their own tools ready
>> yet and are offering C and Linux as a stop-gap measure. But even if
>
> They haven't mentioned 'their own tools' and clearly mention Linux but
> not Windows. They said a design goal was to support Linux so it
> appears
> to be 'their' software from what I read.
That's not what I read. They talk about the "gentle slope" of
*starting* with C, but then later dropping down to the assembler and an
awareness of the underling architecture to take advantage of low-level
details when that kind of performance is needed.
> If you equate C/Linux with enlightenment then yes. ;-)
For the record, I don't. There are applications for which the
combination is fantastic, and others were it makes no sense.
> (You can't have a logical discussion with someone like you who does
> not
> distinguish their religion from their programming. ;-)
I don't know what you're talking about. I'm not just an atheist in
terms of religion, but also an atheist when it comes to programming. I
have no "one path" that I follow or advocate. I pick the one that makes
the most sense. How is that in any sense being religious?
It's *others* (and especially you) who have replaced rationality with
all the trappings of religion in your approach. You've got blind hero
worship, a personality cult, and adherence to a kind of software
fundamentalism that I've always thought was particularly ugly. You've
always had a big problem with me using the metaphor of religion to your
approach. My guess is that you either don't understand the metaphor, or
you are embarrassed by it. Whatever.
>> "this is a Forth chip, you're going to use Forth, period, end of
>> discussion."
>
> IntellaSys says that the native code (assembler) is Forth so whatever
> language you use it will end up being compiled into Forth, end of
> discussion. This is no different than with any other chip, whatever
> language you use on a Pentium it will end up executing Pentium machine
> code. Duh. "C", "J", or even "matlab" will ultimately be generating
> machineForth as the assembler for such a target.
So if I was to write a C compiler for the IntellaSys chips and started
cranking C code at it, you would say I'm programming in Forth? Your
statement doesn't make sense to me.
> Just as a footnote, there are people at IntellaSys programming the
> chips with things other than Forth. IntellaSys has at no time said
> "You are going to use Forth, period, end of discusssion." In fact
> they have talked about use of other common toolchains. So I think
> you are just making up stuff and providing misquotes and more
> misinformation on the subject. (after all you claim to have no
> access to 'real' information. ;-)
I'm not quoting anyone. I'm not providing misinformation. I'm
providing my view, based on what I've seen from what IntellaSys has
released to the public. It's the same information that anyone else can
get themselves-- I have no special access-- and they can come up with
their own view based on the same things.
If you don't like my interpretation, you might consider having your
company doing a better job in terms of documentation.
> But the whole comparison of Tilera and IntellaSys is far worse than
> apples and organges. One is designed for minimal cost and power and
> the other is thousands of times bigger. One has A/D D/A on chip and
> is optimized for realtime while the other has heavy-duty PCI and
> Ethernet and DDR2 hardware interfaces on chip and is optimized for
> Linux.
> One costs $435 in lots of 10,000. So if someone wants to buy a lot
> of
> 10,000 I think there are two or three people in c.l.f who might buy
> one unit from them at cost or accept samples if they are free. ;-)
I'm not sure what you're saying at the end, but I think you've just
restated my point-- that the Tilera and IntellaSys chips are very
different and that to compare them is fruit salad. That won't stop
people in comp.lang.forth from making direct comparisons.
> Both have been shipping in limited quantity. IntellaSys has a
> prioritized
> list including some rather famous companies who's products you all
> love.
I can't wait to hear the details. It's the kind of thing you would
expect would be on the IntellaSys page as examples of design wins.
Geez, who runs your company's marketing?
> If you think your company should be higher on the priority list than
> the
> companies with whom development projects are in the works by all means
> step up to marketing and make a case for how many millions or billions
> of units your company wants.
I don't think my company should be any higher or lower than any other
company interested in IntellaSys products. The other semiconductor
manufacturers we use rarely qualify requests for information with number
of manufactured units. They do when asking for samples or getting price
quotes, but that's not what I'm interested in at this point.
> I should not publish the list or the
> priority but there are certainly people out there working with real
> chips
> even if the hobbiests are not at the top of the list.
That's fine. Your company apparently has limited support and marketing
staff so you have to focus on what is going to give you the most
potential for return. Whatever.
I guess I misread the IntellaSys web site when it suggested that it was
interested in working with users groups, tool builders, and that it
would have evaluation systems available. Now that I know your company
doesn't care about any of that unless high purchase volumes can be
demonstrated, then I'll just look elsewhere for other solutions.
> IntellaSys is certainly not focused on hobbiests or people who want
> zero
> cost systems to play with or who just like to play at writing Forth
> compilers. Unfortunately that is pretty much the profile of most
> c.l.f
> readers and posters. Few work for companies that would even go on
> the
> same list as the leaders in various industries. Fortunately for the
> hobbiests and people who only want free stuff IntellaSys will be
> making
> free hardware available before long.
You may be right about most of the people in comp.lang.forth. I doubt
most hobbyists spend anywhere near the amount we do at work on
evaluation systems and related tools. Too bad my past efforts to get a
simple price quote from IntellaSys on any of the evaluation boards that
have been described fell on deaf ears.
> The Yugo/Ferrari analogy is unreasonable because Tilera is no Ferrari
> and SEAforth is no Yugo. More like jumbo jet and a private jet both
> can fly, both are fast, one is hundreds of times bigger.
You're the one making the association there. I see the applications
that Tilera's chips are perfect for, and I see applications where they
would be grotesque (and expensive) overkill. The same with IntellaSys
chips-- there are some applications we work on where they would probably
be unbeatable, and other applications where they would be a joke.
The Yugo/Ferrari analogy is provided only to illustrate how many people
in comp.lang.forth don't dive into the details and see things as
equivalent when they aren't.
> It is really pretty simple. How about a simple C vs Forth
> comparison? ;-)
Along what axis would you like me to compare them? There are places
where C shines and others were Forth is the best. As I'm not religious
about such things, I don't have a problem of seeing one as always better
than the other. Silly me and my rationality.
> P.S. I know you say you have no access to 'real' information but are
> compelled to post guesses I would suggest that you read the various
> articles in Electronic Products, Network Computing, EE Times,
> Picture Correct, ESCREEN, EE Times Europe, Consumer Electronics Net,
> Business Wire, Circuit Cellar, Forbes, Electronics News, TV
> Technology,
> Engineering Times Asia, Morningstar, TMC Net, Theater at Home, SOC
> Central,
> Computing News, Electronic Design, Media Workstation, Design and
> Reuse,
> ee Product Center, Portable Design, Camera Town, ECN Semiconductor
> Reporter,
> Embedded Technology Journal, EDACafe, SEMISEEK NEWS, DSP Design Line,
> Broadcast Newsroom, DV Format, Dr. Doob's Portal,
Embedded.com, US
> Telecom
> Dailylead All Access, SOC Central, EDN, GG3, eepn, Yahoo Finance,
> digital
> Content Producer .com, HANA, Electronicstalk, Shibui Markets,
>
broadcastbuyer.tv, and the microprocessor report for 'real'
> information
> on IntellaSys.
I do indeed get a large number of those publications and what I see
mentioned about IntellaSys is typically limited to IntellaSys's own
press releases or staff reporters restating press releases. But if you
have examples to the contrary, I'd appreciate either links or if on
paper, publication dates.
Getting back to the strange practices of the marketing staff at your
company, you might tell them to consider actually linking to the actual
articles in question from your web site.
> You could ask for 'real' information in c.l.f or you could ask for and
> post 'unreal' information in c.l.f which is probably a lot more
> fun. ;-)
No, it's not more fun. Despite your characterizations, my interest in
the IntellaSys chips isn't as a hobbyist. The company I work for has
clients with applications where I can see the IntellaSys chips might
have strong advantages. But since apparently they would be dealing in
quantities ranging in the 10k to 100k pieces, that's money that
IntellaSys apparently isn't interested in. Again, whatever.
