VME nostalgia? Heck no. It's really Darwin at work.
Editor’s note: I’m guessing that if you got past the nostalgic title of this column, you’ve probably been around VME long enough to want to reminisce. If not, I’ll be back next month with something more recent: a new title on this book.
It’s VME’s 25th anniversary – or birthday, if you choose to personify VME as a living entity. In many ways, VME is a living entity: We’ve now created a more-than-58-standards progeny in the VITA Standards Organization (VSO), and VME continues to evolve to this day. But, how did we get here? To answer that question, contributing editor Jerry Gipper and I interviewed quite a few old timers – the men and women who were around at VME’s birth.
What surprised me was that few people wanted to reminisce, and even fewer felt nostalgic toward VME. Instead, they were proud of their accomplishments but expressed surprise that VME is still around. Most expected VME to have been replaced by something else by now, having witnessed the rise and ubiquity of the personal computer and its offspring such as PC/104 or other x86-based flavors.
If VME had been a single specification frozen in time, it very well may have gone extinct. But the last 25 years have seen countless changes and modifications to the VMEbus itself and related standards. This Darwin-like continual evolution made VME stronger, and constantly relevant in markets such as test and instrumentation, transportation, and military. Change still sets the stage for VME’s future in mission-critical systems.
Back in the day Jerry Gipper has been around VME nearly since its origin as a Motorola semiconductor marketing strategy used to sell 68000 microprocessors. His Jubilee chronology of VME is worth reading at http://www.vita.com/jubilee/index.html. Our interviews revealed that back then, the 68000, Zilog’s Z8000, Intel’s x86, and several other microprocessors were all vying for market traction. The idea of an amicable collection of competing companies working together to create a board market for 68000 CPUs was highly unusual. This international collection of companies included Motorola’s second-source chip partners Mostek and Signetics, who also threw their weight behind the effort. Undoubtedly, Intel’s then-reputation for being – shall we say, hard to work with – helped the VME effort.
The original VME32 (32-bit) specification evolved to VME64 (64-bit) and eventually to VME64x. The Extensions version of the spec cherry-picked the best attributes of Futurebus, as Wayne Fischer recalled. Even though Futurebus was widely seen as the replacement for VME, loyalists successfully reinvented the spec to today’s most popular version with faster speed, more I/O on P2, and the addition of the P0 connector for user-defined I/O. Eventually concepts such as VME320 and 2eSST would increase the bus speed beyond the original 40 MBps (theoretical) to nearly 320 MBps (pretty close to real world).
VME convinces the Army While VME insiders and their companies were evolving the VME specs, the military was just beginning to flirt with open industry standards in the late ‘80s. When I was working for VISTA Controls, we supplied the Fire Control Processor (FCP) to the U.S. Army’s M1A2 main battle tank. The Army had its own version of an industry standard form factor called SAVA: the Standard Army Vetronics Architecture. Slightly larger than 6U in x and y, with the wedgelocks on the backside, and requiring CDDI (the copper version of the FDDI network) for communication, for a while it looked like VME was going to lose out to SAVA.
Numerous vendors including Aitech, Dy4 Systems, Radstone, and others worked to convince the Army that VME was a better long-term choice and that a program-specific SAVA architecture couldn’t possibly survive (clearly, they were right). I remember years’ worth of extensive deliberations, tests, proposals, and even the engagement of a government civilian contractor (DCS in Washington, DC) before the Army finally accepted the new VME standard as its vetronics baseline. PM Abrams then proceeded to tea bag their VME architecture to other Army platforms including M2A3 (Bradley), Grizzly (Breacher), Heavy Assault Bridge (HAB) and others – giving rise to one of the first instances of COTS reuse anywhere in the DoD.
Another 25 As VME continues to thrive under VITA’s executive director’s vision and a refreshed board of directors, the industry is seeing some big changes. VME industry consolidation continues: Both VISTA Controls and Dy4 Systems, mentioned earlier, are now part of Curtiss-Wright Controls Embedded Computing; the dust is settling on GE Fanuc’s recent acquisitions of Condor Engineering and SBS; and as we went to press, Radstone rebuffed the takeover attempt by Italy’s Eurotech in favor of a bid from GE Fanuc. Through it all, VME is being reinvented yet again as a system-level implementation focused on mission-critical embedded systems.
The industry is now cooperatively creating tomorrow’s VITA 58 “cocoon module”1, and VMEbus Systems magazine will be evolving right alongside it. Starting in December 2006, this magazine will be renamed and expanded to become VME and Critical Systems. Darwin is still at work.