On the Cusp of Visual Computing
As Video Becomes Ubiquitous, The Web Will Need New Infrastructure
By Roger L. Kay
I attended nVidia’s brandy new conference, nVision 2008, Aug. 25-27 in San Jose. And it turned
out to be something that industry conferences are often not: fun. It was fun because it was not
death by slideware, an all-too-common method of torture used by companies on their hapless
victims in the analyst, customer, and media communities. Instead, it was a tour de force of
amazing demos, most of which worked, peppered with celebrity appearances (e.g., Tricia Helfer
of television miniseries Battlestar Galactica, Buzz Aldrin of space mission Apollo 11, space
shuttle commander Eileen Collins, and Nascar driver Kyle Busch).
Since the nature of visual computing is that you can see it, much of what attendees witnessed
was eye candy. There was a 3D animated movie of flies hitching a ride to the moon aboard
Apollo 11, a fantastic display of next-generation touch input by Jeff Han of Perceptive Pixel, and
multiple rooms full of fanatic heavy-duty networked gamers, slugging it out under an energized
emcee’s mike. Oh, and lest I forget, a grand assortment of modified cases was on display,
including one in the shape of a “visible man” medical model.
Yes, there were the tech talk sessions, replete with references to ray tracing, pixel shaders, and
task schedulers. With attendees like this — overwhelmingly male, certifiably geeky, and, in
some (not rare enough) cases, below-par personal hygiene — you need some technical red
meat. Also, for the venture minded there was a more sedate forum, a seven-ring circus of start-
up CEO pitchmeisters speed-dating with the moneyed community.
So, right. Great show. But what struck me was not the show’s individuality, but its emphasis on
a larger theme, which nVidia calls “visual computing.” As a baseline, let me say today’s
computers, at least some of them, can execute spectacular visual tasks. But not all computers
can, and some visual tasks are beyond all but specialized installations. The fact is that the
current infrastructure is not all the way up to handling complex 3D video in real time. First off,
bandwidth must be reliably in the 1Mbps range from one end to the other to deliver high-
definition (HD) streaming. Most consumers in the United States are getting something less.
Video files can run easily into the scores of gigabytes uncompressed, and a couple of weeks
worth of video recording can fill up even the largest hard drives pretty quickly. And even on a
fairly beefy machine in terms of processor and memory, HD video may stutter when run in full-
screen mode.
But as the capabilities of all these pieces of the infrastructure approach readiness for the next
great stage of computing — the video era — a number of companies are lining up to stake their
claims. nVidia, of course, wants to supply graphics processing, arguably the core of the matter,
but so does AMD through its ATI subsidiary, which also supplies graphics technology.
Less obviously, Cisco has a large interest in this area, which it labels “visual networking.” Not
only does Cisco want to sell the higher capability network hardware needed to move all this
visual material around, it also sees a fundamental change in the way computing is done that will
benefit those companies savvy enough to position for it. Rather than moving big video files
around the Web, people are going to want to store it up in the cloud and just pass links around
to their friends and relatives. Viewers are more likely to stream a large file in real time rather
than grab it and bring it down to their home computers, particularly if they plan to watch it only
once or twice. So, most of the heavy duty content will reside in large disk farms tended by video
servers.
Firms positioning for some of the storage action involved in this major shift include EMC, the
traditional leader in storage, and IBM, which has just recently pulled together a major package of
storage products and services for massive installations. EMC refers to its program of managing
large amounts of information rationally as “information management,” a fairly generic term that
nonetheless captures the idea of having to decide how many copies of what to keep where. IBM
places its recent introduction of a product suite of more than 30 hardware, software, and service
elements under the heading “information infrastructure.”
And many other companies, large and small, are focusing furiously on how to position
themselves for the video era. Following a pattern similar to the evolution of the automobile,
computers will get more powerful and complex under the hood, but progressively simpler to use
as the audience morphs from specialist-hobbyists to ordinary people. Early automobile drivers
had to know how to rebuild an engine. Now, all you need to know is how to work a key, a wheel,
a few pedals, and a cup holder, despite the fact that today’s cars outperform anything available
in 1925.
The era of visual computing, visual networking, information management, or information
infrastructure is upon us, whichever term you choose. On the one hand, computing will be more
exciting and fun — and less confusing, too. On the other, individuals and companies will have to
buy a lot of new equipment and services. But who’s still driving that 1925 Model T?
© 2008 Endpoint Technologies Associates, Inc. All rights reserved.
out to be something that industry conferences are often not: fun. It was fun because it was not
death by slideware, an all-too-common method of torture used by companies on their hapless
victims in the analyst, customer, and media communities. Instead, it was a tour de force of
amazing demos, most of which worked, peppered with celebrity appearances (e.g., Tricia Helfer
of television miniseries Battlestar Galactica, Buzz Aldrin of space mission Apollo 11, space
shuttle commander Eileen Collins, and Nascar driver Kyle Busch).
Since the nature of visual computing is that you can see it, much of what attendees witnessed
was eye candy. There was a 3D animated movie of flies hitching a ride to the moon aboard
Apollo 11, a fantastic display of next-generation touch input by Jeff Han of Perceptive Pixel, and
multiple rooms full of fanatic heavy-duty networked gamers, slugging it out under an energized
emcee’s mike. Oh, and lest I forget, a grand assortment of modified cases was on display,
including one in the shape of a “visible man” medical model.
Yes, there were the tech talk sessions, replete with references to ray tracing, pixel shaders, and
task schedulers. With attendees like this — overwhelmingly male, certifiably geeky, and, in
some (not rare enough) cases, below-par personal hygiene — you need some technical red
meat. Also, for the venture minded there was a more sedate forum, a seven-ring circus of start-
up CEO pitchmeisters speed-dating with the moneyed community.
So, right. Great show. But what struck me was not the show’s individuality, but its emphasis on
a larger theme, which nVidia calls “visual computing.” As a baseline, let me say today’s
computers, at least some of them, can execute spectacular visual tasks. But not all computers
can, and some visual tasks are beyond all but specialized installations. The fact is that the
current infrastructure is not all the way up to handling complex 3D video in real time. First off,
bandwidth must be reliably in the 1Mbps range from one end to the other to deliver high-
definition (HD) streaming. Most consumers in the United States are getting something less.
Video files can run easily into the scores of gigabytes uncompressed, and a couple of weeks
worth of video recording can fill up even the largest hard drives pretty quickly. And even on a
fairly beefy machine in terms of processor and memory, HD video may stutter when run in full-
screen mode.
But as the capabilities of all these pieces of the infrastructure approach readiness for the next
great stage of computing — the video era — a number of companies are lining up to stake their
claims. nVidia, of course, wants to supply graphics processing, arguably the core of the matter,
but so does AMD through its ATI subsidiary, which also supplies graphics technology.
Less obviously, Cisco has a large interest in this area, which it labels “visual networking.” Not
only does Cisco want to sell the higher capability network hardware needed to move all this
visual material around, it also sees a fundamental change in the way computing is done that will
benefit those companies savvy enough to position for it. Rather than moving big video files
around the Web, people are going to want to store it up in the cloud and just pass links around
to their friends and relatives. Viewers are more likely to stream a large file in real time rather
than grab it and bring it down to their home computers, particularly if they plan to watch it only
once or twice. So, most of the heavy duty content will reside in large disk farms tended by video
servers.
Firms positioning for some of the storage action involved in this major shift include EMC, the
traditional leader in storage, and IBM, which has just recently pulled together a major package of
storage products and services for massive installations. EMC refers to its program of managing
large amounts of information rationally as “information management,” a fairly generic term that
nonetheless captures the idea of having to decide how many copies of what to keep where. IBM
places its recent introduction of a product suite of more than 30 hardware, software, and service
elements under the heading “information infrastructure.”
And many other companies, large and small, are focusing furiously on how to position
themselves for the video era. Following a pattern similar to the evolution of the automobile,
computers will get more powerful and complex under the hood, but progressively simpler to use
as the audience morphs from specialist-hobbyists to ordinary people. Early automobile drivers
had to know how to rebuild an engine. Now, all you need to know is how to work a key, a wheel,
a few pedals, and a cup holder, despite the fact that today’s cars outperform anything available
in 1925.
The era of visual computing, visual networking, information management, or information
infrastructure is upon us, whichever term you choose. On the one hand, computing will be more
exciting and fun — and less confusing, too. On the other, individuals and companies will have to
buy a lot of new equipment and services. But who’s still driving that 1925 Model T?
© 2008 Endpoint Technologies Associates, Inc. All rights reserved.
Get Ready for the Build-out
 |
| |
 |