Discontinued IC's

From: Chris Kennedy <chris_at_mainecoon.com>
Date: Sun Jul 30 22:39:28 2000

William Donzelli wrote:

> Yes, but you are not the first to be concerned with this - the U.S.
> military was many years ago as many of the parts used in weapons systems
> were being discontinued by U.S. sources. To solve the problem, they
> created VHDL, a complex language used to describe the behavior of the
> parts, from the very high level to the very low. It does not detail any
> of the internal structure of the chips - just behavior.

Well, sort of. Back-annotation is rampant at clock speeds above 150MHz
in both VHDL and Verilog, and you can't build a processor (or something
that looks like a processor) these days without using custom cells,
i.e., non-synthesized hunks of code which describe the circuit
at the transistor level. Purely behavioral code is rarely seen these
days and is often avoided by policy -- it's possible to write perfectly
valid behavioral code which is utterly unsynthisizable -- all of
which makes me wonder how this addresses the concern (if it _is_ a concern).

As for the military -- while the US military loves VHDL (as
do most European firms) most US firms prefer Verilog (IBM is a
notable exception) and with the push towards COTS or repackaged COTS
components the Pentagon is in no position to tell people to write
stuff in VHDL.

Stuff gets obsoleted so fast that the US is busy either preparing to
field (the F22) or compete (the JSF) aircraft which contain
discontinued parts in their avionics systems. They don't fix this
by requiring people to write stuff in VHDL, they fix it by buying
shitloads of parts while they can and then sticking them on the
shelf.

There's also the weird secondary market that exists because of military
programs. TI has previously sold the IP and remaining dies for end-of-life
parts to other firms who then either continue production or package the
dies to meet military needs -- which is great if you need something in
a flat-pac...

-- 
Chris Kennedy
chris_at_mainecoon.com
http://www.mainecoon.com
PGP fingerprint: 4E99 10B6 7253 B048 6685  6CBC 55E1 20A3 108D AB97
Received on Sun Jul 30 2000 - 22:39:28 BST

This archive was generated by hypermail 2.3.0 : Fri Oct 10 2014 - 23:32:58 BST