|
|
 |
Projekte: Archiv
|
Following links will lead you to more project
specific information.
-
We are interested in several important
questions regarding the hierarchical top-down
design of very large circuits. The key aspect
is Top-down Chip Planning based on an
accurate area and timing estimation. A complete
prototype design system, called PLAYOUT,
has been developed to verify our chip planning
ideas. Integrating many tools into a design
environment and performing real large test
designs requires research effort in the areas
of CAD Frameworks (design modelling,
data and design management) as well as Reuse
of Design Objects, too.
-
-
Our seconds research area is the design of
parallel computer architectures for special
application areas. These architectures are used
for verifying our Chip Planning results.
However, we always search for application areas
which are of scientific interest. Until now, we
developed computer architectures in two
application areas: artificial
intelligence and medical volume
visualization.
Our first project lasted from 1985 until 1992.
There, we developed a pipelined multiprocessor
architecture, called PESA, for
accelerating the execution of OPS5 programs. An
extended version of PESA, called PESA-P,
supports Prolog programs, too.
The PESA project has been finished and replaced
by the current VVP project (Volume
Visualization Project). A multiprocessor
architecture, called DIV2A
(Distributed Volume Visualization
Architecture), with self-designed processing
elements is able to perform raytracing of
rotated medical volume data in realtime. The
goal is to compute 25 projections of a
256x256x256 volume (CT or MR data) per second.
In addition to the volume visualization itself
we are also working on volume
segmentation.
 PESA
- Parallel Expert System Architecture
(fertiggestellt)
Volumen-Segmentierung
Fragen und Anregungen zu dieser WWW Seite an:
Bernd Schürmann (schuerma@informatik.uni-kl.de)
|
|
