i can't get the point on both comments, we are all talking on a wintel
perspective of the problem.
actually, generic function CPUs were born as a way to handle different
tasks on the same hardware device.
this would make implementing "especial" features easier since all this
features should be done giving instructions for a single multi purpose
hardware, for sure Software was born.
the problem here is not on the paradigm design, but on the fact that
current microprocessor architectures don't scale.
that is why specialized DSPs are used for, a "Software DSP" is not an
efficient design because of the flaws of current hardware.
BTW, the Oxygen Project is working just on fixing that
(http://www.cag.lcs.mit.edu/raw)
also, an not less important, "Software DSP" are not good on OS Design
perspective, because hardware should be OS-agnostic, if we all want to put
all the work that should be done on "satellite" devices into the main core
of a computer design/OS we are just going to need a bigger (and more
expensive) central processor than we normaly need and a single point of
failure (more probability of system crash)
for this last point, i've found a nicer article on this address
http://www.byte.com/column/Be_View/BYT19990727
Have a nice day
Carlo
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/