"System on a Chip": The 90’s Version
This is one of a series of articles intended to give historical insight on how the workplace has evolved over the last several decades, from a personal perspective. This article covers reflections from a hardware designer's inside view of IBM in the 1980s and 1990s.
Whenever I hear the term ‘system on a chip’ it reminds me of my days at IBM. When I arrived at IBM in the early 1980s, the company culture was still very much embedded with the Thomas J. Watson philosophies, given that either the father (Sr.) or son (Jr.) had led the company from 1914 to 1971. The Watson family values included guaranteeing their employees a job for life. This guarantee meant that whenever a plant closed, or work diminished in one location, everyone was offered the opportunity for re-training and relocation. Over time IBM, the acronym for “International Business Machines” as the company was named by Thomas J. Watson Sr., earned a new meaning. IBM eventually became known throughout the company grapevine as: “I’ve Been Moved”. Yet those of us starting in the company in the 1980s were proud to be called IBMers.
IBM eventually became known throughout the company grapevine as: “I’ve Been Moved”.
Although IBM had introduced the IBM PC by the early 1980s, IBM was still considered the dominant player in the computer mainframe industry. As leaders in the computer domain, IBM seemed impervious to countering trends that were beginning to take shape across the industry. To this end, IBM employees used internal proprietary software to complete their work from general office suite applications to Application Specific Integrated Circuit (ASIC) design development. For example, IBM had its own office suite of products for spreadsheets (Lotus 123) and for creating presentations (Foils5). For presentations, program outputs were printed on sheets of transparency film, called foils. In meetings, the presenter would place each “foil”, one at a time, on an overhead projector, which displayed the image onto a screen.
In a similar way, when it came to ASIC design, hardware design engineers worked on what were called ‘dumb’ terminals’ that hooked to an IBM mainframe, and used IBM proprietary design and development tools, and internally developed physical libraries and physical timing simulators to develop their designs. Also, during this period of time, hardware engineers were assigned to dedicated ASICs. For example, one engineer became the expert on Static Random-Access Memory, another on Floating Point Arithmetic, and another on Control logic and each of these functions were designed and delivered in individual “dedicated function” ASICs. Hardware designers used Karnaugh maps and Boolean logic to minimize the number of logic gates and define how much ‘logic’ could ‘run’ in one clock cycle between two registers. The amount of logic and corresponding functionality that would ‘fit’ between any two registers was determined by the timing requirements of the technology. We were hardware designers, designing ASICs at the register-transfer level of the design.
As time marched forward, this type of IBM culture persisted until the year 1993; that was the year that the culture seemed to change overnight. The IBM Board of Directors decided it was time for change; the board was ready to bring in someone new ‘from the outside’ to run IBM as Chairman of the Board. Early in 1993, for the first time in the history of the company, IBM had a round of layoffs (Ref 1). By the end of the year, IBM hired Lou Gerstner, a new hire from outside the company. The IBM rumor mill reported that Lou Gerstner’s first action as Chairman of the Board was to eliminate the use of ‘foils’ in the executive conference room so that the leadership could meet without being distracted by the art of ‘foil’ making.
The IBM rumor mill reported...Lou Gerstner’s first action as Chairman of the Board was to eliminate the use of ‘foils’ in the executive conference room
At the same time, IBM had been transitioning internally to the use of personal computers, and the hardware design area had transitioned to using a common set of industry available design tools. IBM was reaping the benefits of their involvement and support of a decade long government sponsored Very High-Speed Integrated Circuit (VHSIC) program, whose primary purpose was to advance the state of the art of integrated circuit design (Ref 2). One of the main achievements of the VHSIC program was the development of the VHSIC Hardware Description Language (VHDL) (now IEEE 1076 standard). Associated design automation tools allowed designers to develop increasingly complex designs at a lower cost by seamlessly moving from system concept to system design through a streamlined set of tools, using the VHDL common language. Soon logic synthesis tools were developed that became better over time at translating the ‘coded design’ into a transistor level physical design.
As a result of this whirlwind of changes, the 90s version of the ‘system on a chip’ emerged. By the early 90s, design engineers could design one ASIC containing memory, logic, and state machines - to replace what was previously a set of ASICs. With these advancements in technology and tool development, applications in design re-use led the way to ‘generic’ chip designs. For example, in generic bus interface ASICs, hardware macros could be shared across ASICs guaranteeing consistent system level behavior (see figure at top of article). From our perspective as hardware designers, we became efficient at software programming in order to produce the desired hardware. While the software coding we completed was at a lower level of coding than today’s software programmers, it was far less meticulous than the earlier work at the register-transfer level of the design. We also viewed ourselves as having progressed to the system level of the design. As far as we were concerned, the designs we produced represented the 90’s version of a ‘system on a chip’.
From our perspective as hardware designers, we became efficient at software programming in order to produce...a ‘system on a chip’.
References:
1. First Layoffs Seen at I.B.M., By REUTERS- FEB. 16, 1993, https://www.nytimes.com/1993/02/16/business/first-layoffs-seen-at-ibm.html
2. Very High Speed Integrated Circuits - VHSIC - Final Program Report, 30 Sep 1990, OFFICE OF THE UNDER SECRETARY OF DEFENSE (ACQUISITION) WASHINGTON DC, https://apps.dtic.mil/dtic/tr/fulltext/u2/a230012.pdf