 |
|
DAILY NEWS AND INFORMATION FOR THE GLOBAL GRID COMMUNITY / JUNE 02, 2003: VOL. 2 NO. 22
|
Special Features:
GRID COMPUTING FOR YOU AND ME
By Ahmar Abbas, Managing Director, Grid Technology Partners
One of the first applications that I see coming down the pike that brings grid
computing right into our homes is in the digital entertainment segment. Today,
the digital video camera is one of the fastest selling consumer electronic
products. In fact, Forrester Research estimates that 92% of US households will
be dabbling in digital content creation by 2005.
The proliferation of digital video cameras is, in turn, driving demand in the
consumer sector for digital video manipulation products such as editors,
compositing tools, DVD creation tools and the like. Many of these tools output
video in a number of different formats and thus, feature native video encoding
capabilities.
So, after shooting the future Palme D’Or winning masterpiece with our digital
camera there begins the painful process of encoding the video stream to
compress the raw digital video data in order to make transmission and storage
more practical. Some of the most popular encoding schemes are those developed
by the Moving Pictures Expert Group (MPEG). MPEG-4 is the most recent of the
commonly implemented MPEG standards. Like other encoding schemes, the MPEG4
encoding process is computationally intensive and takes a frustratingly long
time. But should it?
With 31 million homes having more than one PC and half of those homes having
local area networks, it wouldn’t be too outrageous to want to distribute the
encoding process to the other (presumably idle) PCs in our home.
GridIron Software, based in Ottawa, Canada is one of the companies already
thinking along these same lines. They recently announced that they have
successfully encoded MPEG-4 video using multiple computers with its GridIron
XLR8 software for distributed computing.
Taking advantage of the MPEG4 specification which represents content using a
hierarchy of video objects, GridIron was able to treat the encoding process as
an embarrassingly parallel software algorithm with a reasonable level of
granularity to accommodate typical CPU and bandwidth capabilities. Their
published test results indicate that a 60 second NTSC broadcast was compressed
from a raw file size of 923 Mbytes to an MPEG4 encoded file of 13 Mbytes – a
98.75% reduction in size. The encoding time was reduced by a factor of 20 on a
system of 12 nodes and decreased from 2 hours 33 minutes to 7 minutes and 4
seconds.
Most importantly, the test result showed a linear reduction in encoding time.
So even if the video encoding process is distributed amongst just two home
PCs, the encoding time will be reduced by a half.
To create our Personal Digital Entertainment Grid (PDEG) we now have to
convince Sony, Panasonic, Adobe and the like to add this proven distributed
computing capability to their encoding software. Hopefully they are already
paying attention!
Please send comments, questions, shekels and admonishments to
feedback@gridpartners.com.
|