VLSI Birds of a Feather

[PAD Powell[Admin]]

There was a VLSI BOF at the Usenix.  About 1 1/2 hours was devoted to the
discussion of VLSI design software.  The following totally unsubstantiatable
rumours and statements were made.  These are unsubstantiated because nobody
would make them officially,  and we all know the reactions of the various
Blue Suits to the problems.

1.  Bell Labs will never again release any CAD software of any sort for
	any reason.  This is an example of the "Once Burned, Twice Shy"
	syndrome.
2.  The fellows down at MCNC indicated that they would be willing to distribute
	their tools that they have developed.  Copies of the Z paper were
	made available;  further documentation will be posted in either
	a directory on their machine or in net.sources.  General reaction
	of people at the BOF and of me after reading the paper is that the
	package is well thought out and very useful.  They also described
	their language (ABCD: A Better Circuit Description), EFG
	(Editor For Graphics), HIJ, KLM, etc.  One of the few examples of
	humour seen at USENIX.
3.  Various people described what their groups were doing.  University
	of Washington is working on DARPA funded project to gather
	up tools and distribute them to other people.  They indicated that
	there was not much problem gathering,  but distribution is a
	real nightmare.   A project funded by Philips in Europe is
	pounding along, and has some tools available for European Universities.
	Philips is providing a silicon foundary service for the project.
	U. of Waterloo is working on the Canadian Microelectronics
	project, involving the selection of a workstation and developement
	of CAD tools for the Canadian Universities.  This venture also
	involves Queens, where the VLSIIC people are providing foundary
	services for University Courses.
4.  A general cursing session involving the lack of a standard for mask
	generation or description.  There was an expressed desire for a
	published CIF or whatever standard.  Perhaps we should get an IEEE
	committee working on this?  In addition, there was griping about
	the lack of conversion programs from CIF to mask generator format.
5.  People indicated that they were interested in various types of simulators.
	General conclusion is that Unix is not a nice environment to run
	a simulator written in Fortran.  Clearly there is a user community
	here which needs a better fortran under UNIX.  Perhaps, now that DEC
	supports UNIX,  it might be possible to "port" their VMS fortran
	compiler into UNIX.  It might (Heaven Forbid) even become the basis
	for paying money to DEC for a compiler.
6.  There was great interest in the transporting of data/programs between
	various systems.  U. Waterloo experience in IBM/UNIX data transfer
	was recounted;  libel laws do not permit expression of the authors
	views about this experience.   It is clear that a UUCP/UUX
	facility for VMS/UNIX/VM would be really nice, and in fact would
	be essential.  This is really a reinvention of the effort that
	Bell Labs went through with their Designers's Workbench things,
	presented at the 1981 Design Automation Conference.

7.  It was decided to attempt to rename net.lsi into net.vlsi.  This would
	be done by creating net.vlsi,  and terminating net.lsi.  Also,
	people would like to know what happened to fa.vlsi, as nothing
	has been coming over into usenet lately.

It was fun meeting all of you,  and I hope we can do this again.  Remember,
there will be a conference up here in Waterloo in September;  details will
be posted in net.vlsi

Patrick ("Did anybody get any sleep?") Powell