*** IDF Spring 2004
By John N. Latta
February 17 – 19
San Francisco, CA
IDF Spring 2004 #1 – 2/17/04
Welcome to Camp Intel
There is no better an example of Intel being like the military than here at the Intel
Developers Forum. The messaging is scripted. There are no surprises. Reality is only
confirmed if it has been approved in advance. One of the better examples is the
announcement by Craig Barrett, CEO, Intel, of the 64 bit extensions to Xeon. The
reality is that Intel’s hand was forced by the market to extend the 32 bit architecture
because AMD has been successful at doing what Intel has said it would not do. Equally
as important, apparently HP has told Intel it will sell servers using AMD's Opteron. Yet,
today even when asked, Craig Barrett would not concede that the Intel implementation
of the 64 bit extensions will be totally compatible with Opteron. The reality is that the
market responded to a superior solution, over Itanium, and Intel could not ignore it any
longer.
Craig Barrett, CEO, Keynote – Everyone Expected It
So the big news was the 64 bit extension announcement. Done.
Craig also spoke that times have changed in the adoption of technology. In the past,
technology was only made available and the early adopters attached to it. Now the
market must be prepared to absorb the technology. This includes many factors which
are a part of the ecosystem. His point is that technology is no longer created, marketed,
sold and used in isolation.
One of the more interesting demonstrations was UWB as a USB replacement
technology. They claimed the technology was running at 480Mb/s.
In the question session, Craig was most bullish on WiMAX to create a digital broadband
infrastructure. He stated that when it comes to last mile solutions of 50Mb/s and up to
30km, no other technology, at least in Greenfield applications, can match the low
infrastructure cost of wireless. In this context, he stated that the undeveloped countries
are likely to use WiMAX technology earlier and faster than developed countries. This is
consistent with what Sean Maloney had to say at WCA.
Digital Home – Where is the Meat?
Louis Burns, VP & GM, Desktop Platforms Group, showed a television using the Intel
LCOS display chip, discussed Sandow, the 2005 entertainment PC concept, outlined
the DHWG and the NMPR Validation process as two complementary ways to support
CE initiatives. Yet, it was not until the afternoon with the session “Digital Home
Overview and Direction” that many of the weaknesses were evident.
Using a traditional PC industry approach, especially Intel’s, there is a drive to achieve
interoperability. This is rolled into efforts which Intel and many others support. There
are element standards such as 802.11, UPnP AV, CDS, RIO and others to make sure
that equipment works together. On top of this is the Digital Home Working Group which
is focused on interoperability across PC, CE and mobile devices. On top of this is
NMPR (Networked Media Products Requirements) driven by Intel.
Weakness:
This is driven by companies that want a share of the CE market. As a result they have
packed their bags just as they always have: standards, working groups, interoperability
and commoditization.
It is not at all clear, other than from a supply side, that such an approach is embraced
by consumers and that it is, a priori, necessary. Today no CE company sells its
products based on interoperability with other CE companies. Yes, there are some
standards such as NTSC video, a.k.a. RS-170, but these are largely signal interface
standards.
The interoperability framework was presented at 6 layers: DRM, Media Formats, Media
Transport, Device Discovery and Control, Network Protocol and Physical Network. This
was claimed to be between a PC and a media server. Yet, IPv6 was not included and
there is no solution to the DRM issue. All DRM proposals, to date, have been
proprietary and it is unclear that any one company seeks a single vendor solution.
Weakness:
To say that there are many DRM approaches and the consumer would not know in
advance what they are, and that they are not interoperable, is a solution seeking market
implosion. Unless the DRM solution is interoperable there is no interoperability at the
system level. In fact, having a proprietary DRM solution is going down the path that the
CE industry has today – they seek customers to buy all components from one
company.
Compelling Content is the foundation for the digital home and those enabling content
guidelines are due out later this year. Examples include: Music on Demand, Creativity
(whatever this means), Video on Demand, PVR/EPG, Hi-DEF, and games. This can be
viewed at 1’, 2’ and 10’, local or remote, scheduled on the web and targeted content
edited and lastly sent via e-mail, web, disc, or DMA.
Weaknesses:
This thread ran throughout the presentations today - any content, any device, any
location. One can only concluded that we all must be standing around or sitting down,
all day, digesting content. This is the media ubiquity model taken to the extreme. Or, put
in another way, media myopia. Surely there is more going on in the home that just
watching television or listing to music [that is a pretty dumb statement]. But Intel, like
many other companies, is chasing this.
The Digital Home Ecosystem Stack includes: Media processing silicon, middlewear,
system integrators, ODMs, PC applications and Online Content Services. Not a single
company of the 31 shown are CE companies.
Weaknesses:
This is the wantabee crowd – like, “being on the Intel train gives us a chance in the CE
market.” One must keep in mind that Intel’s approach to standards and interoperability
benefits Intel more than any other company. When it is all driven to commodity,
especially at the silicon level, those with market dominance are in the strongest position
to reap the rewards.
But there are other shortfalls in these presentations. It goes back to what Craig said in
his keynote about the market being prepared to accept technology and the ecosystem
around it. Missing here is any discussion of:
Role of brand name in CE,
Marketing of products,
Production and scale,
Distribution,
Price points and price elasticity,
Retail, and
Customer support.
On the CE side one has to only look to Sony, Panasonic, Philips, RCA and Samsung to
see what is required. For the U.S. to reenter this market is not new. Think of
TiVO,
RePlay,
Ceiva,
Audrey,
Kerbango, and
Intel microscope
as examples of how difficult this is.
If there was any question that this is a PC-based event, the Technology Showcase, i.e.,
exhibit floor, made that obvious. Not a CE device to be seen.
For the PC industry to make substantial changes to the CE industry, or to even become
a major factor in it, is much deeper than working groups, silicon that does media well or
middleware. It is about a major and sustained commitment to a well-established market
and ecosystem. What we have seen at CES and now at IDF is a long way away from
that commitment.
The March of Processors
Bill Siu, VP & GM, Desktop Platforms Group, and Mike Fister, Senior VP & GM,
Enterprise Platforms Group, gave a keynote on the End-to-End Enterprise. 2003 was
called the Milestone Year for Itanium 2. It is claimed to be #1 in the performance
benchmarks; there are 50+ systems shipping and 1,000 applications have been ported
to it. The cheers were much more muted when it was announced that only 110,000
have been shipped. No wonder Mike Fister is struggling to meet quota. Six new code
name processors were shown on the roadmap. Enhanced processor technologies
coming include: virtualization, power management, I/O & Storage, Memory and
reliability. It was implied in 5 years that the low end of the Itanium would be at price
parity with the high end Xeon. More was made of the 64 bit extension technology. They
even claimed the company has been working on this a long time. Shane Robison, Chief
Strategy & Technology Officer, HP, sung praises for the 64 bit extensions but said
nothing about their intent to ship AMD Opteron systems.
Anand Chandrasekher, VP and GM, Mobil Platforms Group, showed design concepts
for the 2005 mobile PC called Florence. There are three versions: Florence 12” - On the
Go, 15” - Virtual Office, and 17” - Mobil Entertainment PC. The latter got the most
attention. The keyboard slips from the integrated case and is wireless. On the left side
is a VoIP phone which pops from housing and on the right is a media PC remote. What
is striking is that this could become a Longhorn Media PC reference design and that the
overall ID is quite distinct from what we see today in either notebooks or media PCs.
Anand also outlined major activities in the mobil sector. Coming in H2 2004 is Sonoma
which combines the Dothan chip with the Alviso chip set. This will support 802.11 a/b/g.
An interesting twist - they are suggesting that displays should also be on the outside
case of the notebooks. This can provide useful information including: WLAN signal
strength, battery status, e-mail status and time. There is an interesting comparison to
the sub displays on the outside of cell phone that have gone over very well, especially
in Korea. Note that many of these displays are OLED. Certainly the most exciting
demos were with the Florence notebook designs outlined above.
Digital Home Security
Craig Barrett said it best, the media box is just a PC masking as CE. Yet, the intended
market is CE and we should consider it in this context. The use paradigms for each is
quite different.
PC CE
Turn on Boots Plays
Control Keyboard/Mouse On/Off/Buttons or Hand
Held Control
Connectivity Network/Internet None
Media Software, CD or Broadcast, cable, DVD,
Network CD others
Experience Use Result Media play, replay and
Observation
Failures Based on Type Take to Shop;return
(Some skills product or discard
required)
Update Via Software Almost Never
Security Issues Frequent Never
From a market standpoint the PC media box in the digital home should:
Look like CE,
Bought and used just like CE,
User I/F is CE-like,
Integrate with other CE in the home, and
Not require any attention beyond what is done with CE, i.e., virtually none.
All of these attributes are consistent with the presentations at CES and here.
It is not necessary that this box be connected to the Internet but it is likely that many will
be. Users may get e-mail and browse the web with a small or big screen experience.
Further, when it is connected to the Internet and the home is networked, the PC media
box may be the central device in that network – so called media server. In this
environment the perceived value of the PC media box is increased. In spite of all of
these factors, the success of this home appliance is that it is a CE duck – walks like CE
and quacks like CE and in the consumer mind, it is CE.
The problem is that it is a sitting CE security duck.
If one can accept the analogy, there are parallels with the attacks on 9/11. In the pre-
9/11 environment we had grown complacent with security and, with relatively small
efforts, terrorists could cause significant long-term damage and change to society. Such
is the impact of exploiting holes in the security fabric of society.
In the PC world, we are in the middle of a terror campaign. Spam is threatening the
value of e-mail, spyware eats at the fabric of the privacy of the computer, others may
download executable programs on a PC with ease and it is possible, without the user’s
knowledge, to turn a PC into a spam engine or DOS attack instrument.
What then does this have to do with the CE box a consumer buys? It is a PC under the
skin but not seen as such by the consumer. A million or more home networks that have
an “old” PC, in a security context, at the center, is a prime terror target. The market
impact of one malicious virus campaign against these PC media boxes could literally kill
the market for them. As virus attacks get more sophisticated, it is not unreasonable to
expect that specific targets emerge. As with MyDoom, that target was SCO, but it could
just as easily be 1m+ digitally networked homes.
The potential for attack is in fact amplified by how the box is used in a network context.
That is, it may be connected directly to the Internet with no firewall, it is not a managed
box or network, the means of browsing and gaining e-mail varies significantly between
consumers and it can be subject to attacks when found on the Internet.
This takes us back to the user paradigms shown above. The CE use model and
consumer expectations do not relate to security or updating. Ask the consumer if they
have updated a Panasonic VTR or Sony home media center and only stares result. To
expect that the user will update a PC media box is unrealistic – this is antithetical to the
use expectations. As a result, it is highly likely that over time home networks will
become increasingly vulnerable to attack. The attackers may gain satisfaction from the
result but the market result will be disastrous.
How to avoid this? There are many creative ways. Some include: hosted ISP services,
mandatory PC media box upgrading, consumer firewalls, hardware protections,
remotely managed services and many more. Virtually all of these either translate into
the consumers spending more money, at the time of purchase or on a monthly basis,
or, with rising use, complexity or both. Neither one is fitting within the user CE
expectations. Placing a PC media box in the home cannot result in the home becoming
an IT bastion, if so it defeats the whole purpose of entering this market. Thus, a priori,
from a supplier perspective, it is impossible to predict what systems in the field, over
time, will be vulnerable and, thus, impossible to estimate the impact of targeted attacks.
The end result, managing assault situations is very difficult, if not impossible.
What does all of this have to do with IDF? Quite simply since the issue was not being
discussed, the WAVE Report asked questions. First in the Digital Home Experience
area the following was asked: “How will the consumer digital home network be
protected?” The quick answer was: DTCP-IP. Yet, this is for the protection of media
rights from abuse by consumers, not for the protection of the consumer’s computer.
Yes, was the response to my observation.
Then at the session on an Opt-in Strategy for a Safer Computing Platform, the WAVE
Report asked if TPM could be used to protect the consumer. TPM was clearly not
designed for this; it would only provide some protection and certainly not for the
variability of viruses being seen today.
Thus, the responses, to this limited probing were insightful and reinforced the concern.
It appears that the effort at protecting the consumer investment in the digital home
pales in comparison to that invested in DRM. Thus, when the supplier of content is
more important than the buyer of the home infrastructure, only trouble looms. The CE
duck is sitting.
Wireless USB is hot but keep in mind it is just another name for an UWB application.
NEC was showing a demonstration on the Showcase floor. Volume production is
expected in 2005. On the technology road, Intel has shown it can play standards politics
as well as the best. With the IEEE 802.15.3a standards effort log jammed between the
OFDM (Intel) and DS-CDMA (Motorola) camps, Intel walked with TI and will use the
MBOA (Multiband OFDM Alliance) as the means to complete the MAC and PHY
standard. This effectively neuters the IEEE activities. So much for all the charts that
Intel has been giving on the rapid ramp impact that IEEE standards have. The message
is, this rapid ramp rule only applies when Intel agrees with the standard being
proposed.
In his keynote, Pat Gelsinger, CTO, Intel, spoke of the Era of Tera. He showed a
300mm silicon wafer which has multiple radios on a chip. The key to this design is a
reconfigurable architecture which includes a 4X4X4 Reconfigurable PLA, a 256-state
Viterbi ACS and 16 X 15b Multiplier. This is beginning to approach the architecture
required to implement the Software Defined Radios shown at the SDR Forum. These
were state machines implemented in FPGAs. Intel has yet to speak about its plans in
cognitive radios.
In the Intel military, Moore’s Law is the doctrine. Pat began his talk by asking – is
Moore’s Law at the end of the road? Not hard to guess the answer to that question. But
in the process, Pat described many of the challenges which lie ahead. His glimpse of
the future was centered on digital immersion. Digital should be a part of our lives every
minute, everywhere and with everyone. This creates massive quantities of information
and Intel’s expects this demand to drive the demand for processing. He then described
the RMS view which has the components of: Recognition, Mining and Synthesis. This
perspective is totally information centric and largely visual in framework. Pat is
projecting that our lives will be transformed by these massive quantities of information.
What was glaringly missing in Pat’s keynote was any concept of action or work. By
implication, all human acts are cerebral, which seems naively incomplete. In fact, most
of what we do in a day is based on action or work. Driving an automobile uses many
forms of work and mechanical advantage which has the driver in control. Human
movement is a form of action and the most consistent passive form of human activity is
sleep.
The logical response to the question of how can technology help. Man cannot focus on
just information. The major gains from technology in the last 100 years were in power
and transportation which are either forms of action and work or enable them. Robotics
is a technical implementation of action and work as is a dishwasher. As we look to the
future of what technology can do for man, it includes: biotechnology, information and
action/work.
Wireless USB – Trying to Make it Just Like USB.
With surprising speed the first UWB application has surfaced – wireless USB. Intel has
taken this beyond USB and seeks to provide a transport which will support 1394,
Bluetooth and USB 2.0. They are seeking to use a convergence layer above the UWB
MAC which will support all three of these protocols. They call this WiMEDIA.
All of these activities did not converge until February 2004, that is, they were
announced yesterday, 2/17. The promoter companies are Agere Systems, HP, Intel,
Microsoft, NEC, Philips and Samsung. The requirements include:
480Mb/s bandwidth to 3 meters
Up to 1Gb/s
Power management including sleep/listen and wake
Security
Ease of Use which parallels USB
Bridge to wired USB devices
Host to Device wireless connectivity
Cluster on a Host of up to 127 devices
QoS support
The specification from MBOA covering the MAC and PHY layer are due Q2 2004. V 1.0
of the Wireless USB specification is due early 2005. First standards based products are
to surface in mid-2005.
In the NEC booth was a demonstration of Wireless USB. Its silicon partner is Staccato
Communications. The demo uses a Xilinx FPGA but the final silicon will be an ASIC. It
was stated by NEC that the receiver would be direct conversion. Engineering samples
of the silicon are due at the end of 2004. Production in 2005. The products will use the
full UWB spectrum – 3.1 to 4.9 GHz.
Wave Issue 0411 3/26/04 Article 1-01