From 1983 to1984, RUBICAD’S president and founder Michael Reinhardt was involved in a university research project to explore the use of Hardware Description Languages (HDLs) in developing complex IC designs. At the time, CALMA workstations were considered very expensive and high performance equipment, and DAISY Systems had just released its first x86-based workstations for schematic entry and logic simulation. At this time Michael, together with five colleagues, developed a 32-bit microprocessor in 2.0-micron technology. They implemented the processor on different levels, from the HDL, register transfer, and gate level down to the GDSII layout level. The layout was done with a lot of manual work and many night shifts on a 68000 microprocessor-based workstation with university layout software. 

After 18 months of hard work, the final design contained two ALUs, stack cache, and four independent, asynchronous working control units for I/O, instruction decoding, and operand fetch and execution. It contained about 120,000 transistors on a die of 11 mm x 11 mm. Unfortunately, the university fab line could only produce dies up to 6 mm x 6 mm at that time, about half the die size needed for the group’s design. 

Way Back in the Mid-80s 
It looks tiny and easy to implement today, but in 1983 the best-performing commercially available microprocessor was the 16-bit 68000 from Motorola, which contained approximately 68,000 transistors. There was no 32-bit microprocessor on the market at that time. 

At the end of the project, the group went to an IC manufacturer to present their work. There they learned that the manufacturer also used a 2.0-micron technology, but that, thanks to a better lithography technology, all of this technology’s contact rules and metal rules were much smaller than they were in the group’s design, so small that the group’s this design could be easily produced in the manufacturer’s technology. 

An Idea was Born 
This was the moment when a discussion began within the group about the possibility and usefulness of a program that could convert designs from one technology to another. 
A short time later in his professional consulting career, Michael Reinhardt faced the same problem again. At that time, the average design was smaller than 30,000 transistors, but was being produced with an entirely manual layout. For this reason, many IC companies were very interested in compaction technology for process migration and for achieving higher densities.

Michael initiated and lead a research group combining the forces of academia and industry to study and solve the problem. The group discovered that the existing approaches didn't solve the 45-degree problem, delivered insufficient layout quality, and used far too many resources in terms of run time and memory. Therefore, they developed a new compaction algorithm based on a scan line algorithm.

Many Nights Spent at the Computer
After six years of research, the first prototypes were ready, which delivered good results compared to other algorithms. The tool was used for several commercial ICs and delivered working silicon every time. 
Just then, the progress in process technology speeded up and three- and four-metal layer processes made automatic place and route the technology of choice. 
But Michael foresaw other applications for the technology. From his background in computer science, he knew that the only solution to developing large systems within a reasonable time frame was the method we call today "design reuse". His logical thought process was: either there is a limit to continued shrinks in technology dimensions, or the circuits will become so complex in terms of transistor count that the only solution besides full automatic design is design reuse.

Today we know that both of these conclusions are true: the circuits have become too complex to design from scratch, and there are limitations to the linear reduction of technology dimension, as well as the fact that methods such as linear shrink, automatic place & route, and synthesis have their limitations.

So Michael founded RUBICAD and built a new compaction system called LACE from scratch. It now has an easy-to-use GUI and many features that weren't available in the original prototype version.

It Was Only the Beginning 
The research didn't stop there.: it wasonly the beginning. Since the company’s founding, RUBICAD has added a great deal of additional technology features to the initial LACE product. These  include layout editing, mask operation, partitioning, complex design rules, and hierarchy. 
Most of these innovations have improved  the tool’s output: the quality of design the tool produces has improved dramatically. 
Many groups in the industry which have been working on these different problems separately--such as timing, signal integrity, power consumption, metal migration, phase shift mask and OPC--are now looking to layout manipulation and compaction technology as a solution to all of them.

Back to Top