18.099 contributing to the Grid

                Humanist Discussion Group, Vol. 18, No. 99.
       Centre for Computing in the Humanities, King's College London
                   www.kcl.ac.uk/humanities/cch/humanist/
                        www.princeton.edu/humanist/
                     Submit to: humanist_at_princeton.edu

         Date: Thu, 29 Jul 2004 08:05:45 +0100
         From: Baden Hughes <badenh_at_cs.mu.OZ.AU>
         Subject: Re: 18.095 contributing to the Grid?

> Date: Mon, 26 Jul 2004 09:49:13 +0100
> From: Willard McCarty <willard.mccarty_at_kcl.ac.uk>
> >
> Recently I was made aware of a scheme for harnessing the power of idle
> machines, grid.org. "Grid.org is a single destination site for large-scale
> research projects powered by the United Devices grid computing solution,
> <http://www.ud.com/solutions/mp_global.htm>Grid MP Global. With the
> participation of over 2 million devices worldwide, grid.org projects like
> <http://grid.org//projects/cancer/>Cancer Research,
> <http://grid.org//projects/anthrax/>Anthrax Research, and the new
> <http://grid.org//projects/smallpox/>Smallpox Research Project have
> achieved record levels of processing speed and success." One downloads a
> small agent that then takes the unused cycles of one's machine for working
> on small segments of very large problems. This agent turns itself off and
> on again and runs without affecting one's ordinary uses of a machine. On a
> laptop, for example, it turns itself off whenever the machine is on
> battery-power, on again when the machine is connected to the mains.
>
> The scheme seems brilliant and a very fine idea. I would be most grateful
> for any words to the contrary from those who know more about what grid.org
> actually does.

The grid.org model is only one example of the emergence of large scale,
commodity grade, distributed computational services. There are a large
class of scientific and other experiments which are either
sufficiently computationally complex or data-intensive to vastly surpass
the computational capacity of even high-end computational facilities
available to single researchers. In this context, "utility computing" is
the advocated approach - harnessing largely idle computing workstations
through the use of a common middleware solution to compute problems in a
massively parallel manner.

John Nerbonne mentions this as a factor in his ACH/ALLC plenary this year.
I recently talked about the use of such techniques at the Computing Arts
2004 Conference in Newcastle. Humanities researchers are beginning to
approach the bounds of computational efficiency in a range of areas, and
hence consideration of new paradigms of computation is warranted. I know
of humanities grounded research in the areas of sociology, geography, art
history, linguistics and musicology which are using similar types of
technologies to enable the consideration of much larger scale problems that
were
previously able to be efficiently explored.

Regards

Baden
Received on Thu Jul 29 2004 - 03:34:33 EDT