
VESA MEDIA CHANNEL BACKGROUNDER

The Growth of PC Video

In recent years, video has become available on the desktop PC.  Computer
based applications such as interactive training/education, presentations,
video conferencing, video editing and video delivery systems have fuelled
the need for video-enabled PCs.  Simultaneously, PCs have been improving
their price/performance ratio. In particular, the graphics capabilities in terms
of resolution, true color support and refresh rates have improved at an
astonishing rate.

These improvements are making the PC a more suitable platform for video
than ever before.  The availability of many peripherals and equipment such
as video cameras, video digitizers, electronic still cameras, VCR's, and CD-
ROM's has added to the technological trends that accelerate the much
heralded multimedia revolution.

However, the architecture of personal computers has not yet caught up with
the new demands of video processing. Many of the video cards available
today use the VGA Feature Connector, which was originally introduced by
IBM to enable 8514/A passthrough.  The VGA feature connector has since
been standardized by the Video Electronics Standards Association (VESA)
for ISA graphics adapters.

There are a number of reasons for which the VGA feature connector is not
the ideal interface for the transport of video data.  Its range of colors is
limited and its data rate is restricted to the aging 60Hz VGA refresh rate. The
video stream is not really integrated into the computer environment as it
bypasses the computer to appear on the screen. The VGA Feature
Connector approach also requires hardware duplications of frame buffers
and in some cases DACs (Digital to Analog Converters). All of these issues
make it clear that this method of handling video is becoming increasingly
inappropriate.

The need for a Videographics standard

A standardized architecture for proper integration of video streams with the
computer system that overcomes the shortcomings highlighted in the
previous section has been required for some time.

This architecture needs to take into account the very demanding nature of
the video data types, both in bandwidth and processing requirements within
the PC environment.  To achieve high quality video display, the computer
must have the bandwidth to handle the video stream and the ability to
capture in real-time at full frame rate.  This is the latency requirement.

The older computer system buses, such as ISA or MCA have neither the
bandwidth, nor the real-time characteristics needed to deal with
uncompressed video streams. The more recent system buses, such as VL or
PCI, have improved the bandwidth capability but have not addressed the
equally important latency requirements.  Also, most PCs have add-in cards
for networks, graphics, sound and other capabilities, all of which add to the
demands on the system.

All of these issues suggest that the new architecture must incorporate the
use of a dedicated highway with well-defined real-time behavior and
sufficient flexibility for video and streams.  This highway must interconnect
between all subsystems that involve the transaction of uncompressed video
data streams. These subsystems include, but are not limited to, graphics,
video in, video out, video compression and decompression (MPEG, Px64)
and more.





The VESA Media Channel

VESA has defined a new architecture known as the VESA Media Channel
(VMChannel) to address the issues highlighted in Section 1 and to overcome
the problems highlighted in Section 2. This new architecture introduces a
new, open standard for videographics.

The VMChannel architecture addresses the following objectives

o       Cost effective and seamless integration of video and
graphics within a PC.

o       Guaranteed real-time behavior.

o       Minimization of pin/signal requirement.

o       Scalability for complexity and cost/performance trade-offs.

o       Modularity and flexibility to cater for differing system
requirements & upgradability.

o       A technology independent and future-proof approach for
multimedia integration.

Cost effective and seamless integration of video and graphics

To achieve cost effectiveness and transparent integration of video and
graphics, the VMChannel uses a single frame buffer approach in the basic
architecture.  The VMChannel allows video data to feed directly into the
conventional graphics frame buffer.  Therefore, the video streams are
transported at their native rate regardless of the resolution or refresh rate of
the system on which they are displayed and without the need for additional
frame buffers.

Based on a single frame buffer architecture, the VMChannel intentionally
avoids directly using the graphics memory interface in its specification --
unlike some other approaches.  There are two reasons for this choice.  One
is that the graphics memory subsystem is highly dependent on its
configuration. Therefore, any specification dependency on the memory
interface will effect technology independence and future-proof requirements.
The second reason has been the need to maintain the scalability of the
interface.

Guaranteed real-time behavior

Within a VMChannel environment, real time data such as video is
transported as streams.  Streams are packets of data with a header
(identifier) that specifies a particular stream.  The contents of a stream (the
pixel format and associated scaling, etc.) are included in the stream ID used
by a source or destination to identify the stream.  This approach avoids the
need for transferring vast amounts of attributes with each stream and offers
a flexible and open way of handling or extending the system capabilities.  In
particular, other real-time streams such as audio and/or even timing tokens
can be transported within the VMChannel framework.

Minimization of pin/signal requirements

As part of the drive to minimize pin and signal usage, the specification of the
VMChannel uses novel and efficient schemes to cater for the large number
of signals and pins needed for subsidiary functions such as interrupts.   To
avoid using dedicated pins for initial selection and initialization for each
device, VMChannel uses a two wire daisy-chain. This scheme, particularly in
the stand alone case, provides a plug-and-play approach while consuming
very few signals/pins.  Similarly, the channel scheduling and arbitration
mechanism do not require dedicated signals/pins while catering for the real-
time requirements in a video environment.

Modularity and flexibility to cater for different system requirements

One of the key features of the VMChannel is that it was designed to be
compatible with the widest range of graphics controllers and devices such as
video digitizer cards and codecs. This is particularly important because it
allows users to select devices at any level of cost/performance they require.
This is possible because the VMChannel can dynamically resize the bus
width, allowing low cost 8-bit devices to talk transparently to 16 and 32 bit
devices.

Scalability for complexity and cost/performance trade-offs

The VMChannel allows a high level of flexibility and cost engineering in any
one system. This is because it can combine a number of devices such as
video digitizers and compression and decompression systems in a modular
fashion.  For example, a base system - a graphics system supporting the
VMChannel - can easily and cost effectively be configured as a capture,
decode only, encode only or a full encode/decode video system.
Furthermore, a system based on VMChannel can cater to more than one
video stream transaction between video subsystems, or between video
subsystems and graphics subsystem.  Applications such as video-
conferencing require this flexibility as they need to support multiple video
windows.

Another benefit of the VM Channel's scalability is that the customer has the
choice to extend the system beyond the baseline specification.  While this
baseline specification provides a high degree of functionality and
capabilities,  enhancements on scheduling, standalone operation, and
stream attributes can be catered for if required.

A technology-independent and future proof approach for multimedia

The key objective in defining the scheduling mechanism has been to specify
an adequate and simple scheme in the baseline specification while allowing
for well-defined paths for extensions.  The base scheduling scheme can be
simple because the VMChannel is a dedicated real-time highway and does
not suffer from complexities associated with a system bus.  While the base
scheduling mechanism is static, the scheduling command extensions can be
used to optionally implement sophisticated dynamic arbitration.

The basic scheduling of stream transaction activities on the VMChannel is
based on ensuring guaranteed access for a specific time by each agent on
the Channel.  Each device is programmed via a register (grant time register)
to use the channel for no more than a specific time before it has to hand
over the channel to the next device.  Furthermore, the handover mechanism
allows flexibility in the selection of the next device.  This "constrained"
scheduling scheme provides a simple, powerful and extendible mechanism
for channel sharing while not requiring dedicated arbitration signals.

An ideal architecture for PC multimedia integration

Finally, the VMChannel has been specified to avoid direct dependency on
any particular technology and to ensure future-proof characteristics.  The
real-time behavior, modularity, flexibility and scalability of the VMChannel
make an ideal and truly open architecture for PC multimedia integration.

Vesa Media Channel
White paper
Mel Jackson
VideoLogic
October 1993

































