Well, we're not quite there yet, though that fellow (what's his name . .
. ), the CEO of Oracle, is promoting a situation in which you'd not buy
software at all but would simply search for what you wanted, buy it "by the
drink" over the net, and run it once, repeating the process, and paying for
it each time you use it. I certainly hope that such a scenario is not where
we're headed.
The technology has been in front of us for some time, and what's prevented
this so far is the FPGA vendors' tight hold on how to configure their
products. Once that cat's out of the bag, someone will surely try this,
since he can then incorporate his configuration generator into his software
hierarchy. That will certainly put an end to the question of which CPU runs
which application the best, since the application will then be integrated
with the configuration of hardware which suits it the best.
Dick
-----Original Message-----
From: John Foust <jfoust(a)threedee.com>
To: Discussion re-collecting of classic computers
<classiccmp(a)u.washington.edu>
Date: Friday, August 20, 1999 12:45 PM
Subject: Re: Article on old software programs (for IBM, Apple, Borland, etc)
At 11:04 AM 8/20/99 -0600, Richard Erlacher wrote:
>
>I don't think we'll have to wait long for a situation to arise, in which
one
>links to a site on the web, is fed a download of
configuration data which
>defines how the system on which one's running is to be defined, then reads
>the code which will be executed on the specifically configured "hardware"
>environment. Instead of the hardware defining how the software must be
>configured, the software will define the way in which the hardware
addresses
its
requirements.
It's already here. The Transmeta CPU could work that way.
Field-programmable gate arrays (Xilinx chips) have existed for
many years, so who knows how they're being used.
For example, the circa 1992 Video Toaster for the Amiga relied
heavily on these chips to perform real-time video manipulations.
Load a new transition, reprogram the array.
- John