CPLD fusemap data - why the secrecy?

I have been using the Coolruner for some time now.
I use tools which I have written so I needed the
programming data; back then, Philips gave me some of it
(and what they did not give me - the ZIA multiplexing
truth tables - I reverse engineered myself).
When Xilinx took over, I tried to get such data for their
newer parts, only to be told that I have to buy
a $9 million worth parts per 3 months to be entitled
to ask about these data. I did not bother asking after exactly
which 3 months (the first or perhaps the tenth) I would be allowed
to ask for the data again.

 Here comes the question to the group: has anyone ever been
successful getting fusemap data from Xilinx? (under NDA
or whatever).
 Another (less vital, more out of curiousity) question:
Why is the secrecy? Can anyone suggest some other plausible
reason for it except the obvious (control of their customers)?

Please save the "why do you need it" and "they offer free tools"
kinds of answers, I am aware of all this.

Regards,

Dimiter

-------------------------------------------------------
Dimiter Popoff               Transgalactic Instruments
http://www.tgisci.com
-------------------------------------------------------

dp@tgi-sci.com wrote:

>  Here comes the question to the group: has anyone ever been
> successful getting fusemap data from Xilinx? (under NDA
> or whatever).

I don't write synthesis tools so
I would have no use for such data.

>  Another (less vital, more out of curiousity) question:
> Why is the secrecy? Can anyone suggest some other plausible
> reason for it except the obvious (control of their customers)?

Sure. Not having to publish, verify, update and explain
a formal external document for every rev of every part.

Customers have too many options to be controlled in any case.

    -- Mike Treseler

On 18 Jun 2005 09:08:28 -0700, "dp@tgi-sci.com" <dp@tgi-sci.com> wrote:

>I have been using the Coolruner for some time now.
>I use tools which I have written so I needed the
>programming data; back then, Philips gave me some of it
>(and what they did not give me - the ZIA multiplexing
>truth tables - I reverse engineered myself).
>When Xilinx took over, I tried to get such data for their
>newer parts, only to be told that I have to buy
>a $9 million worth parts per 3 months to be entitled
>to ask about these data. I did not bother asking after exactly
>which 3 months (the first or perhaps the tenth) I would be allowed
>to ask for the data again.
>
> Here comes the question to the group: has anyone ever been
>successful getting fusemap data from Xilinx? (under NDA
>or whatever).
> Another (less vital, more out of curiousity) question:
>Why is the secrecy? Can anyone suggest some other plausible
>reason for it except the obvious (control of their customers)?

It costs time and money to provide this documentation in an end-user consumable form, and there
would be little if any commercial benefit to them in doing it. 

If they had a policy of publishing a spec then it would make it harder for them to make changes to
the silicon as the changes would need to be documented as well. If they have a limited number of
people who need to know the details, then it's much easier (i.e. cheaper) to make sure everyone is
up to date.  

Documentation may reveal details of the architecture etc. that they regard as trade secrets. It may
also reveal possible patent infringements. 

So the bottom line is that it's just easier and cheaper for them not to publically document the
internal details that the vast majority of customers have no need to know about. 
 

>It costs time and money to provide this documentation in an end-user >consumable form, and there
>would be little if any commercial benefit to them in doing it.

Actually they have to document it properly either way.
Many people work on such projects, how can you write all the software
it takes without documentation. The question is, why is it kept secret.
I am not asking for any explanations/support, I can make my way
through the data.

>Documentation may reveal details of the architecture etc. that they regard as >trade secrets. It may
>also reveal possible patent infringements

That's their public version, I know. Have you ever seen such data which
does reveal any "trade secrets"? Nothing of the kind in the data I have
been using so far.

Dimiter

--------------------------------------------------------------------------------
Dimiter Popoff                        Transgalactic Instrumets
http://www.tgi-sci.com
--------------------------------------------------------------------------------

dp@tgi-sci.com wrote:

> Actually they have to document it properly either way.

That's not always the case.
Internal specs can be much more informal than external ones.
I have done NDAs involving only sketchy
text documents and indirect email Q and A.
Technical guys actually doing the work are
spread pretty thin.

             -- Mike Treseler

>That's not always the case.
>Internal specs can be much more informal than external ones.

Well I would be happy with these informal data as well - under NDA
or whatever. I never asked for more.

Notice that this still does not explain the secrecy.

They wanted me to make a $9 million quarterly revenue so I would
be eligible to sign an NDA for these data (and they never said they
did not have them).

Dimiter
------------------------------=AD------------------------------------------
Dimiter Popoff               Transgalactic Instruments
http://www.tgi-sci.com
------------------------------=AD------------------------------------------

Why these endless arguments? Dimitr, just read the postings in this
thread, slowly and carefully.
Documenting and maintaining documentation in publicly usable form is
expensive. And the avarage user has no need to know the things you ask
for, and would not even know what to do with the information.
Those are two overwhelming reasons not to make the data public.
PLD companies try to make money, and avoid waste.
Now, if you are such a good customer that you sepnd $ 9M in a quesrter,
we will be extra nice to you...

Why is this difficult to understand?
Peter Alfke

Peter Alfke,

are you sure you read my postings?

>Documenting and maintaining documentation in publicly usable form is
>expensive

Like I explained, I would be happy with the data as they are - under
NDA
or whatever. I made this clear from the very start and I did get the
data
from Philips, only Xilinx are secretive.

> And the avarage user has no need to know the things you ask
>for, and would not even know what to do with the information

How often do average users ask for fusemap data under NDA
and are denied it? How many of your average users do successfully
use their internally written software to program Coolrunners based
on what they could achieve before Xilinx took over and blocked it all?

>Those are two overwhelming reasons not to make the data public

Which of them do you think applies to my postings so far?

>>Why is this difficult to understand?

Dimiter
------------------------------=AD=AD-----------------------------=AD-------=
------

Dimiter Popoff               Transgalactic Instruments
http://www.tgi-sci.com
------------------------------=AD=AD-----------------------------=AD-------=
------

dp@tgi-sci.com wrote:
> Peter Alfke,
> 
> are you sure you read my postings?
> 
> 
>>Documenting and maintaining documentation in publicly usable form is
>>expensive
> 
> 
> Like I explained, I would be happy with the data as they are - under
> NDA or whatever. I made this clear from the very start and I did get the
> data from Philips, only Xilinx are secretive.

So we know Philips <> Xilinx.

>>And the avarage user has no need to know the things you ask
>>for, and would not even know what to do with the information
> 
> 
> How often do average users ask for fusemap data under NDA
> and are denied it? How many of your average users do successfully
> use their internally written software to program Coolrunners based
> on what they could achieve before Xilinx took over and blocked it all?
> 
> 
>>Those are two overwhelming reasons not to make the data public
> 
> 
> Which of them do you think applies to my postings so far?

Probably both ?
[ and more -> what happens, if you believe the Data <> fuse ?
- support bandwidth is needed ]

Perhaps you should explain WHY their present tool flow is not good 
enough for your task, and how your flow is much better ?

This is a serious point - you may well have soemthing Xilinx have not
thought of, that could benefit everyone.

There ARE intermediate formats, that enable you to get very close to
the iron, and you then work only with their fitter portion of the tools.

Last time I checked, the coolrunner flows were happy with TT2 and BLIF
files. Have you looked at that ?

-jg

>or whatever. I made this clear from the very start and I did get the
>data
>from Philips, only Xilinx are secretive.

Xilinx is still in the (C)PLD business.  Philips isn't.  Maybe
that's because Philips didn't make the right decisions on things
like this.

Even though you may not need any support, it's hard to determine
that ahead of time.  If they give you the info and you can't understand
it or it's buggy or ..., then you will be even more pissed off at them
than you are now.

If they give you the info, then the next guy who isn't so smart will
expect to get it too.


Somebody at Xilinx thinks that support or whatever would be too expensive
for the amount of money they will get in return.

How many $ do you spend on their chips?  Do you have any alternatives?
How much more $ will they get from you if they give you the info you
want?  ...  Is it even worth it for them to think about the problem?

I'm not sure what you want to do.  As Jim Granville suggested,
my guess is your best approach is to make your tools feed the
back end of Xilinx's tool set.

-- 
The suespammers.org mail server is located in California.  So are all my
other mailboxes.  Please do not send unsolicited bulk e-mail or unsolicited
commercial e-mail to my suespammers.org address or any of my other addresses.
These are my opinions, not necessarily my employer's.  I hate spam.