Jeez, wait 'till Mr. H. gets a load of your posts....
Mr. H has a sense of humor.. It's those dapper Alex's (Delay and Bullcansay) who take themselves too seriously.
Can you give a comparison of ZiVA and Impact as a DVD solution?
The first is really architected for settop boxes where the
CPU is realtively weak (runs Windows CE) and the display is
always NTSC or PAL out, whereas the second is a broader chip
meant for PCs where CPU power (Windows NT) is a freebie and
the presentation format (display) can be *anything*.
Fixed function (fomer) is expected for the box, but the PC is
a very fluid place where DVD will be constantly changing shape:
display modes, coded sample rates, drivers, graphics,
available memory, optional DVD features that are later enabled,
user interfaces/navigators, private data streams, even whole
new ideo and audio types, new applications for DVD, and so
forth. The hardware designers will go crazy trying to keep up,
fighting bugs (as they already are).
Do you know what I think most DVD decoders both on the PC
and settop will look like in the long term? A: devices like
Cyrix' MXi! (they even had the intelligence to add a byte
average instruction, something Intel's brain-dead MMX
completely missed).
On high end PCs, it will be graphics controllers that
significantly accelerate or take on the entire MPEG process.
There WILL BE high end 3-D chips which can decode most
stages of HDTV (High Level) by '99 and inexpensive dedicated
single-chip HDTV decoders too!
How do you rate the offerings from all the single
Chip encoder suppliers?
It's getting harder to wade through just the ones I know
about, let alone the ones I don't. I would imagine about
10 people at Cube think they can go off and start their
own pet encoder chip company, plus 50 more people out
there in the Collective MPEG Implementation Land.
Formula: Give yourself and a few of your buddies a million
shares each, start an HDL sweat shop, slap a bunch of modules together, tape out, and call it an encoder! Then pray to get
bought out as quickly as possible. How boring. Few if any
could really pull it off. Architect cockiness/igno-arrogance
again (NOTE: you WANT your competition to be like this!).
Players:
Cube slow but sure. DVx Video quality: highest,
but most expensive. Later and less flexible and more difficult
to program than Kohnhead had hoped but still the best
architecture (no surprise) in this category of dedicated
encoders. The bonus 2xML+ decode capability may win friends
at Microsoft (McMahon and Fate). The apparently "smaller"
motion range here isn't really an issue (people who think
search range is the important thing remind me of people who think size is everything---not what you do with it). Strong
pre-processor.
I fear Cube will become complacent and charge higher prices
(this is another motive for the hoards to do their own),
aiming for the high end too long. It would be nice to have
this in CE units.
Sony The CXD1922Q is a sincere chip that will make
its way into medium cost ($500) consumer electronics devices
in 2H 1999 when after is has been shrunk to 0.25 micron. OK
video quality. It has problems with scene changes (fades)
reminiscent of Cube's E1->E3 series. Slow to integrate.
There really isn't enough push yet to put it into systems.
Unlike DVx, it won't be aimed at teleconferencing (low_delay)
since it only deals with frame structured pictures.
IBM cost between Sony and Cube. Image quality: better
than Sony, but not as good as Cube's. Much faster to integrate
into ystems thanks to 3-chip and MCM history. More product
focus and neither IBM nor Sony need the galleon of 40+
microcoders like Cube.
Innovacomedy
late. Quality ?!? Chez-D ?
CRisky and any others you care to include
Won't make it. Employee ownership (15%) too low.
No real video experts at helm. So-long Chin buyout
too expensive. Back to drawing board.
Others:
Hitachi--original MPEGCam was an MPEG-2 product, but had power
problems. MPEG-2 proto in late '98 (?).
NEC: still having a hard time programming their encoder.
Panasonic's (http://www.sscs.org/isscc/1998/ap/TP2.htm)
single chip is late, but again is sincere. Panasonic even
had a substitute 15-chip encoder card at Comdex just to show
they care about DVD authoring.
Zapex(Xing): is integrating their PGA/DSP bank into a single(?)
Chip, although they have traditionally aimed at the high-end
broadcast market.
VisionTech/DML: don't know anything about them yet.
There a plenty of customer-jokers like Optibate, Vitepid
who think they are doing their own. Lotsa luck.
Who does LU?is LU (illucidt ?).
Why after years of trying will 80&T Microelectronics
finally get it right ?
The encoder market will someday be very much like the decoder
market which is already becoming like the graphics controller
market: "nobody else is doing one, lets do it!", then everybody
and their motherboard, followed by a shake-out, and a finally
handful of cutthroat survivors.
CHROMAC -- The Omniscient |
