The Cyber-Spy.Com Usenet Archive Feeds Directly
From The Open And Publicly Available Newsgroup
This Group And Thousands Of Others Are Available
On Most IS NNTP News Servers On Port 119.
Cyber-Spy.Com Is NOT Responsible For Any Topic,
Opinions Or Content Posted To This Or Any Other
Newsgroup. This Web Archive Of The Newsgroup And
Posts Are For Informational Purposes Only.
From: Ian Montgomerie
Subject: Re: Open Source Consumer Products
Date: Fri, 22 Nov 2002 09:55:52 -0800
References: <firstname.lastname@example.org> <email@example.com> <firstname.lastname@example.org> <email@example.com> <firstname.lastname@example.org>
NNTP-Posting-Host: 12-235-107-216.client.attbi.com (220.127.116.11)
X-Newsreader: Forte Agent 1.9/32.560
On Thu, 21 Nov 2002 10:34:42 +0100, "Geraldo S."
>"Lewin A.R.W. Edwards" wrote in message
>> > What's the rationale behind this story then? A VCR is a pretty one-sided
>> All consumer products are built with custom parts.
>That's simply not true. At best *some* high volume consumer products contain
>custom parts. And most IC's in VCR's can be purchased on the world market so
>they're not 'custom parts' anyway. Most smaller manufacturers (such as Quad,
>Linn, Mark Levinson) use ONLY off-the-shelf parts because they don't have
>the resources to design their own IC's. Even Sony is moving more and more
>towards COTS parts, such as the C-CUBE MPEG2 decoders in their DVD players
>and the Tripath digital amplifier chips in their DVD-receivers.
The two of you are using different definitions of "custom parts".
These ASSPs (like the Ziva MPEG decoders) are very customized to do
their specific job, which is why they're "application specific". The
fact that they are sold to lots of people does not mean that they have
any common architecture with anything else. For example, the Ziva
MPEG decoder does not contain a general purpose microprocessor (DSP or
otherwise) that you could write C code for. It uses an arcane
microcode, and the thought of open sourcing this is highly amusing.
ASIC and ASSP chips are best regarded as "custom parts" even if sold
to many different manufacturers. They are VERY different beasts from
standard CPUs and DSPs.
>> It will be cheaper, for the foreseeable future, to use ASICs and
>> ASSPs, than general-purpose devices. If it wasn't so, people would be
>> using GP devices universally already because of the huge savings in
>> training engineers to work with generation n+1 of parts.
>And yet they are taking that road, as more and more functionality is
>contained in the software of the appliance.
This is only partially true. For typical consumer products, the
natural road of advancement is toward cheaper chips that do more.
When combining many fuctions in one chip, typically more leeway must
be given to software control code and microcode to do different
things. But, for example, anything that processes compressed
audio/video data needs a substantial amount of custom acceleration and
processing to do its job with a minimum amount of silicon. And the
amount of custom acceleration/processing is closely tuned to the jobs
it is supposed to do. So for example, adapting an MPEG2 video codec
chip to work effectively with MPEG4 at similar resolutions is
typically impossible, without significant hardware changes.
>> Upgradability is a positive NON-SELLING point to manufacturers. An
>> ASIC-based product for $30 or a high-end-DSP-based product for $90?
>> What GUARANTEE do you have that it will even be possible to upgrade
>> the $90 product to a future standard? What if today's flash media are
>> superseded by 3" DVD-RAM disks? Even if the $90 device has the
>> horsepower to decode the new format, it doesn't have the other
>> necessary hardware. You simply cannot future-proof a CE product.
>> Trying to do so, beyond some some very tight constraints, is expensive
>> and pointless. A good case in point is some [brand] modems that were
>> sold as being software upgradable for "any future standard" (at a time
>> when V.32bis was state of the art). V.32terbo was OK, just a firmware
>> upgrade. Well, along came V.FC and they discover that the DSP is too
>> slow and the DAA misconfigured for this protocol. So it WASN'T
>> upgradable for "any future standard".
>For every example you can give, I can cite a dozen examples where software
>upgradeabillity has helped the product stave off obsolesence.
How about just one example from _consumer electronics_.
>And there's another reason: software upgradeabillity will allow
>manufacturers to fix mistakes in their products once they're on the market.
This is a false argument. Flash updates to fix bugs have absolutely
nothing to do with making a product using an ASIC or ASSP rather than
a standard DSP. Significant silicon bugs in shipping consumer
products are actually quite rare. Approximately all bugs are caused
by microcode or control code. The ability to change the code to fix
bugs (by physically replacing a ROM chip if necessary) is quite
distinct from the ability to add new features.
Go Back To The Cyber-Spy.Com
Usenet Web Archive Index Of
The sci.electronics.design Newsgroup