edited and compiled by Sam Pettus (aka "the Scribe")
with assistance from the many fans and followers of the Internet emuscene
a series of articles focusing on the birth and growth of computer system emulation from 1964 to 1998
copyright © 1999 Zophar's Domain,
all rights reserved
Sharing the software
IBM and the very first computer emulator
by Sam Pettus
It all started
with a problem.
Most inventions come about that way. Thomas Edison, the great American inventor, is often quoted as saying, "Necessity is the mother of invention." He has been proven right more times than one could possibly count over the march of time, and his maxim holds true even when new technology comes along.
IBM had a problem. They needed to solve it, fast.
Emulation was invented to solve that problem.
Now we are stuck with it. There is no going back to the days without emulation. Never again.
ruled the American computer and data processing industries from the end
of World War II to the early 1980s. Everybody followed their lead,
made their stuff to work with their products, or made their stuff to work
just like their products. They were to the mainframe computer industry
of that time what Microsoft is to the personal computer industry today.
IBM set the standard. If it wasn't IBM compatible, or at the least
IBM inspired, then it was either quite unique or quietly ignored by many
government and business organizations. Even today, the phrase "IBM
compatable" remains with us long after their dominant days are over, and
the concepts that underlie that phrase were set during this time.
At the close of the 1950s, mainframe computers such as the IBM 1410 and the IBM 7070 ruled the industry. True, there were offerings by other vendors such as Digital, Sperry-Rand, Honeywell, Burroughs, and so on, but everybody tended to follow the trends that IBM set. Not everybody was happy with this situation, though, with many of them offering machines that were faster, better clones of existing IBM products. IBM was understandably upset at this, so in order to maintain their technological lead they initated a study known today as the SPREAD report. One of the key items in that report was that it called for something that was quite novel at the time.
Remember, these were the early years of mainframe computers. These systems were designed as a whole, and there was little, if any, component sharing among them, not to mention program porting. Each unique system design required unique software for proper use, and the lack of cross-platform capability among its older product lines was a fact that had finally dawned on IBM's R&D teams. This fact was all the more emphasized just a few years later in 1962, when a report came in that "an anonymous customer" had hardware-hacked their IBM 705 so that it could run programs written for the IBM 1410. They decided that any future IBM systems needed to be back-compatable with their older product lines. IBM's mainframes were the number one choice of the bulk of the growing computer market. If IBM wanted to keep that market share, then newer systems would need to support the software for older ones, at least until replacement software for the newer systems was developed. Providing this capability would give more incentive to prospective buyers of the NPL (New Product Line, as internal IBM memoranda constantly referenced it), and that would in time result in increased sales. Old IBM products ruled the day, but that day would in time draw to its close. Newer, more advanced IBM products would have to be able to work like the old ones if IBM were to continue its reign as top dog in the computer industry.
Thus it was that in late September of 1963, IBM commissioned its World Trade Corporation Laboratory in La Gaude, France, to develop a series of simulation programs designed to mimic the performance of its seven most popular mainframe computers at the time. These programs were done as a test to see how the NPL prototypes could provide back-compatability based on a pure software solution alone. The results, when they came back, were rather discouraging. The most effective simulation ran twice as slow as the actual machine, and the worst simulation ran just over ten times slower than its inspiration. It was clear to IBM executives that a back-compatability solution based entirely in software was an unsatisfactory one. It would be far too slow for mainstream system use. Another solution had to be found in order to make back-compatability an integral part of the NPL. Later that year, while the NPL systems were still in development, IBM executives assigned systems engineer Stuart Tucker the thorny problem of finding a back-compatability solution for the NPL. He had just over a year to do it.
In the meantime, though, IBM's competitors weren't going to just wait around for the NPL to arrive. Honeywell released its own mainframe, the H-200, that could run IBM programs. This drove home the need for IBM to include cross-platform capability in the NPL, but something had to be done about Honeywell, and quickly. John Haanstra, another IBM engineer and one not overly fond of the NPL project, proposed a stopgap measure that would offer cross-platform capability entirely via hardware. Market need meant that Haanstra's proposal was quickly approved, and the result was the IBM 1410S mainframe. Still, it was quite obvious to IBM that solutions such as Haanstra's were only temporary ones at best. Another approach was needed, and it had to be one that was not limited to the inflexibility of a pure hardware approach.
One of the members of Tucker's staff at IBM's Poughkeepsie, NJ development center (located in the United States) was a rising young talent by the name of Larry Moss. It was his idea that the pure software approach used by the La Gaude teams was wrong, as the hardware and software of the day was too limited to handle such a sophisticated simulation. The closer that the designed hardware came to resembling the target system, then the better such an approach would work. He proposed an approach that combined both hardware and software elements in such a way as to provide support for the IBM 7070, one of the most sophisticated IBM machines then available. This approach would go beyond a mere simulation; it would involve a real product working in conjunction with an NPL computer that would result in a machine that could successfully execute IBM 7070 programs. This product strove to be like a true IBM 7070, or to put it another way it attempted to emulate its performance by running the same software, but on a different machine. Moss used the word emulate to distinguish his proposal from the La Gaude simulations due to its dictionary definition, "to strive to be like," and he called his proposed product an emulator. Tucker liked the proposal and approved it, and the rest is history.
The NPL project was commercially released in June 1965 as the IBM System/360 family of mainframe computers, which set a new standard of technological excellence for their time. The name System/360 came from the geometric expression "360 degrees," intimating a well-rounded computer system, and the slash was added to emphasize that it would be a complete break from earlier IBM product lines. Along with the System/360 product line came the Moss emulator, which was more commonly known under the name 7070 Emulator. Larry Moss conceived it, Stuart Tucker approved it, and fellow engineer Joe Brown is credited with actually building the working prototype. It proved to be a big hit with IBM customers, and many of them were running their old IBM 7070 programs on the System/360s well into the late 1960s and early 1970s. It never bothered them, and they never worried about the fact that their business-critical applications were now being handed by a new and unique technology known to us today as computer emulation ...
... and now ...
you know ... the rest of the story. Good day!
Pugh, Emerson W., Johnson, Lyle R., and Palmer, John H. "Solving the Conversion Problem." IBM's 360 and Early 370 Systems (Cambridge: MIT Press, 1991), pp. 157 - 164, 168, 702.
Pugh, Emerson W. "The Road to System/360." Memories That Shaped an Industry (Cambridge: MIT Press, 1984), pp. 203-204, 229, 292.
Photos courtesy of the University
of Kentucky Computer Sciences Department and Lincoln Land Community College.
If you would like to learn more about the IBM System/360, then see what
IBM has to
say about their history-making accomplishment!