Project Title:
--------------
Benchmarking and analysis of architectures for CAD applications
Abstract:
-----------

Computer aided design tools play an important role in design and
manufacture of today's VLSI systems. The performance of these tools
critically depend on the architecture of the machines on which they
are run. Hence it is important for the VLSI designers to know which
machines are good for their applications.  There is a need to have a
suite of benchmarks which reflects the performance of different
machines on CAD applications. However, the current standard SPEC92
benchmark set contains CAD applications which are completely outdated
in the sense that they are no longer representative of applications in
CAD domain.

The goal of this work are:
i) Create a benchmark suite consisting of CAD tools representative of
the workload in today's design methodology,
ii) Benchmark a collection of state-of-the-art architectures from
various computer vendors on these applications,
iii) A thorough investigatation of memory hierarchy effects on the
performance of machines on CAD applications


Work Division:
--------------
The three parts of the project can be easily divided into parallel
workloads. 
i) Each of us will be responsible for the compilation of a set of CAD
applications.  
ii) The set of machines to be benchmarked will be divided amongst us. 
iii) We intend to investigate the effect of configuration and
management of cache and the main memory on the performance of CAD applications. 
Reading Material:
------------------


Computing Resources:
--------------------

We would need access to SPARC (with sunOS) and HP machines. Also, we
would need to use Dinero (cache simulator) in the third part of the
project. For a better analysis of modern cache design (with
interleaved access for improving memory bandwidth, reference: project
P10 in CS252 Spring 94), we would like a more sophisticated cache
simulator.  If you have any suggestions, please let us know.

Accomplishment till 18th March:
------------------------------
i)Collection of source codes for all the applications to be used in
the benchmark set.

ii) Compilation of at least half the applications on all the machines.