Sorry for the delay in answering, but...duty calls.
There are not many designs in .25u because most fabs are still
coming up on the technology. Big houses like IBM, Moto, Intel
are in production, but the push through the .35 era was quicker
than expected. Processes with feature sizes smaller than .25u
are still pretty much lab fabrication, not manufacturing.
The paradigm shift in design methodology has got a lot to do with
how the industry is virtualizing itself. Please refer to one of
my earlier appends on the subject. It is more economical for larger
companies to hire a group like IMS to do certain specific work than
it is to do it inhouse. Also, since design methodologies have
essentially converged on a few standards, such as Synopsis, Cadence,
Avant!, whether formal or informal, it is easier to reuse designs.
The analysis and verification programs available today have capabilities to handle huge problem sizes. In the past, for
instance, the analysis might have been of a block or two on a chip,
where now all the blocks on the chip can be analyzed together. This
is all leading to a major increase in individual design productivity,
making possible things like SOC.
The question always was, if you are going to put a system on a chip,
what should the system be? Well, obviously, you would pick the one
that would sell the most copies, and it would become a standard
part. Now, or in the very near future, it will be possible to build
a SOC for niche products, with very quick turn around time, and low
development expense. I expect NSM to be strong in this area, to the
detriment of Intel, for instance. Scotland is warping their whole
advanced education system in order to become a force in SOC. There
are two main groups here, the developers of the IP and the people
that put it together into SOC. The system developers can be a couple of hackers that have a great idea, and if they can get some VC
support they can come to a company like IMS and we will do the
physical design for them, or provide the library, or do verification
work. Whatever it takes to pay the mortgage. The work can be spread
around, but the great networking available today removes a lot of
the distance factors. Bottom line, though, is the innovators don't
have to assemble a design team, get the tools and equipment, and
wait a year or so for the group to start working as a team.
It is a real democratization of the industry,and clearly it is more fun to do this type of work than working for some old line company.
Frank |
