Xilinx Spartan3: Price

Hi to all,
I want to use a Xilinx Spartan3  XC3S1000 FG456 in my new design
Does anybody knows the price (>1000p) and the availability for that chip?
I found a price of about $200 at AVNET
Xilinx is telling a much lower price...?
So what is the truth
Is the chip already available and at what realistic price?

they are both correct.. check the volumes field on the Xilinx web site.. its
probably 250,000

"itsme" <itsme@gmx.de> wrote in message
news:bno138$fok$01$1@news.t-online.com...
> Hi to all,
> I want to use a Xilinx Spartan3  XC3S1000 FG456 in my new design
> Does anybody knows the price (>1000p) and the availability for that chip?
> I found a price of about $200 at AVNET
> Xilinx is telling a much lower price...?
> So what is the truth
> Is the chip already available and at what realistic price?
>
>

"Simon Peacock" <nowhere@to.be.found> writes:

> they are both correct.. check the volumes field on the Xilinx web site.. its
> probably 250,000

Yes. Take a look at the * at the bottom of this page:

http://www.xilinx.com/prs_rls/silicon_spart/03142s3_pricing.htm

Well, if he can team up with 250 of us and make an order of 1000 each
as a single group we could get the $12 (I guess that is the lowest
speed grade in the cheapest packet). 

Petter
-- 
A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing on usenet and in e-mail?

Petter Gustad wrote:
> Well, if he can team up with 250 of us and make an order of 1000 each
> as a single group we could get the $12 (I guess that is the lowest
> speed grade in the cheapest packet).
> 
Just some perspective:
When I joined Xilinx 16 years ago, the price per LUT was about $1.00
Today it is hundred times lower, at around  $ 0.01
And we are promising another factor 10 in the near future. Not too bad !
BlockRAMs and multipliers and DCMs are thrown in "for free".

BTW, the 3S1000 has about 16 000 LUTs. You do the math.

Peter Alfke

Petter Gustad <newsmailcomp6@gustad.com> wrote:
: "Simon Peacock" <nowhere@to.be.found> writes:

:> they are both correct.. check the volumes field on the Xilinx web site.. its
:> probably 250,000

: Yes. Take a look at the * at the bottom of this page:

: http://www.xilinx.com/prs_rls/silicon_spart/03142s3_pricing.htm

: Well, if he can team up with 250 of us and make an order of 1000 each
: as a single group we could get the $12 (I guess that is the lowest
: speed grade in the cheapest packet). 

If everyone of this team of 250 could contribute one day of lead time we
could also get  in a time frame where the parts are really available on the
market :-)
-- 
Uwe Bonnes                bon@elektron.ikp.physik.tu-darmstadt.de

Institut fuer Kernphysik  Schlossgartenstrasse 9  64289 Darmstadt
--------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------

Uwe Bonnes wrote: 
> If everyone of this team of 250 could contribute one day of lead time we
> could also get  in a time frame where the parts are really available on the
> market :-)
> 
One minute per team member would be sufficient.  :-)
Peter Alfke

"Peter Alfke" <peter@xilinx.com> wrote in message
news:3FA00CB5.1766F7F7@xilinx.com...

> Petter Gustad wrote:
> > Well, if he can team up with 250 of us and make an order of 1000 each
> > as a single group we could get the $12 (I guess that is the lowest
> > speed grade in the cheapest packet).
> >
> Just some perspective:
> When I joined Xilinx 16 years ago, the price per LUT was about $1.00
> Today it is hundred times lower, at around  $ 0.01
> And we are promising another factor 10 in the near future. Not too bad !
> BlockRAMs and multipliers and DCMs are thrown in "for free".
>
> BTW, the 3S1000 has about 16 000 LUTs. You do the math.

I Just wonder why the price drops so much between 250k and 1k.

1k  @$200 = $200k
250k  @$ 12 = $3,000k

So he really only needs to find 16 or more people that want 1k to get a
discount, and as a bonus he gets many thousands of free chips.

if he could just find 25 people they would all be buying their 1k @ $120
each and getting 9k free chips.

Why not $50 @ 1k?

Ralph

"Ralph Mason" <masonralph_at_yahoo_dot_com@thisisnotarealaddress.com> writes:

> if he could just find 25 people they would all be buying their 1k @ $120
> each and getting 9k free chips.

Now I'm starting to understand why I'm such a lousy businessman :-(

Petter
-- 
A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing on usenet and in e-mail?

Ralph,

It is also called "forward pricing" because it lets a customer know what it
will cost when they finally go into production a year or so from now.

A real customer doesn't care what it costs right now, because 'right now' is
not when they are going to sell anything.  They want to know what it will cost
when they go into production (along with all of their competitors).  Until
then, they need to budget for the costs, but they are not at all as concerned
about the short terem, as they are about when it is selling, and they are
trying to squeeze the best out of the (potentially thin) margins.

For example, if I know I want to hit the market next Christmas with a new GPS
videogame/handheld/FRS personnal communicator/802.11g (call it Internet
Enabled Star Treck Cache Dragon Hunt*), and I wish to sell 250K units for
$29.95, I need to make the BOM today, get pricing locked in, and be sure that
when I go to build them, I can turn a profit.

The one off price today, or next month is of no value to anyone who is really
in business (other than to let you know how much to write the check for).

Austin


*Game is not patented, or registered, but since I just made it public domain,
I do expect to see three versions of it by next fall.  The unit would
automatically register you on a website as a player, find other players in
your area, assign local FRS channels and tones to the players as teams ("red
vs. blue"), and then send GPS coordinates for "caches" that must be reached
before certain time limits in order to score points.  Once you have arrived at
a cache point, the unit verifies its location, and gets going you off to the
next one.


Ralph Mason wrote:

> "Peter Alfke" <peter@xilinx.com> wrote in message
> news:3FA00CB5.1766F7F7@xilinx.com...
>
> > Petter Gustad wrote:
> > > Well, if he can team up with 250 of us and make an order of 1000 each
> > > as a single group we could get the $12 (I guess that is the lowest
> > > speed grade in the cheapest packet).
> > >
> > Just some perspective:
> > When I joined Xilinx 16 years ago, the price per LUT was about $1.00
> > Today it is hundred times lower, at around  $ 0.01
> > And we are promising another factor 10 in the near future. Not too bad !
> > BlockRAMs and multipliers and DCMs are thrown in "for free".
> >
> > BTW, the 3S1000 has about 16 000 LUTs. You do the math.
>
> I Just wonder why the price drops so much between 250k and 1k.
>
> 1k  @$200 = $200k
> 250k  @$ 12 = $3,000k
>
> So he really only needs to find 16 or more people that want 1k to get a
> discount, and as a bonus he gets many thousands of free chips.
>
> if he could just find 25 people they would all be buying their 1k @ $120
> each and getting 9k free chips.
>
> Why not $50 @ 1k?
>
> Ralph

"Austin Lesea" wrote:

> A real customer doesn't care what it costs right now, because 'right now'
is
> not when they are going to sell anything.  They want to know what it will
cost
> when they go into production (along with all of their competitors).

Nitpicking, but, companies like mine are "real customers" yet we only need a
few hundred or thousand devices for production ... not hundreds of
thousands.

I've always thought that semiconductor pricing was unfairly skewed to favor
the big guys.  I understand discount structures, etc., but there's a huge
difference between $200 and $12.  And, it's weird, 'cause you'd think that
you'd sell a ton more chips if the little guys could buy them at a more
affordable price point.

I've been dealing with a semiconductor manufacturer that's been downright
rude about discounts because they were the only game in town.  Now, with
Virtex 2 Pro's high speed serial I/O capabilities I have a chance to drop
them like a hot potato in my next design.  The likelyhood of that happening
is extremely high at this point.  I'm sure they bend over backwards for
those who move more chips, but, what they don't realize (even though I've
explained it), is that I'm ramping up.  They are literally handing Xilinx
business to the tune of thousands of V2P's per year.  Very unwise.

So, the high cost of chips for the sub 1K/year crowd might very well make
them look elsewhere and, as you can imagine, once you adopt and get
comfortable with another vendor the chances of getting a different chip onto
that a board are pretty low.


-- 
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Martin Euredjian

To send private email:
0_0_0_0_@pacbell.net
where
"0_0_0_0_"  =  "martineu"