More about Onyx Systems

Onyx Systems was founded in 1979 in Cupertino, California by Bob Marsh and other Zilog UNIX enthusiasts.  Venture funding was by Carl Berg, a real estate developer in Silicon Valley, as the means to create a market for 8" hard drives made by another of his technology companies called IMI. The bootstrap system was multi-user Z80 CPM/MPM/PICK system which contained an IMI 10Mbyte and 20Mbyte disks and a 3M 1/4" DC-300 tape cartridge in a large (and heavy) aluminum desktop or rack mount case.

The Onyx C8000 system started with the bootstrap system as an I/O controller and added a Zilog Z8002 16-bit segmented processor with 256K of memory running an almost stock PDP-11 V7 UNIX ported to the Zilog Z8002 instruction set. The system had about the same speed and system architecture as a PDP-11/70 at a fraction of the cost (about $20K in comparison to several hundred thousand dollars). With a 16 segment split I/D 16bit MMU the basic architecture was similar to PDP-11's which made a lot of application and kernel porting relatively straight forward.

The development systems used pre-production silicon, that did not stabilize until just about the same time as we started shipping systems. Usable Motorola 68000 silicon was still months away.

After the initial porting team defaulted in early April 1980, the project was brought in-house at a remote location. After several months of 100-150hr weeks and a strict Keep-It-Simple-Stupid (KISS) development strategy and some couple hundred plus compiler bugs later V7 UNIX was running on the Z8002. The port at this point did not support floating point, had only 60 some lines of V7 C code altered, all new machine support assembly code and drivers. A few days after it responded to the first shell prompt, the applications were compiled, debugged, and we took five of the systems to the NCC computer show (*THE* computer show at that time), several software vendors ported their tools and applications in the motel rooms, and all ran without crashing for the show. A few weeks later, and less than a couple dozen lines of performance tuning changes, we shipped the first units to barely meet the contractual commitments with the Venture Capitalists.

After NCC we took machines to USENIX held in Boulder, CO that year with a huge success. We also gave a machine with sources to UC Berkeley which became an undergraduate staple under the efforts of Erik Fair and others.

A few months later, the VC's flushed the management and UNIX development team with harsh criticism about attempting to create a UNIX market.  Their demand was to focus on more CPM/MPM systems to meet the demand. A year later IBM introduced the IBM PC running PC DOS (MS DOS) and in the first six months sold more microprocessor based systems than had been produced by all vendors combined in the preceding 5 years. The demand for CPM/MPM systems dried up almost instantaneously. Save the several thousand Z8000 UNIX  Onyx managed to sell into AT&T Long Lines, the telco operating companies, and a large number of shcools the company would have been instant toast.

At this point, Onyx had sold more unix machines than all other vendors combined. Usenix, which was a Source License only club dominated University Computer Facility Managers, was pretty anti-business and wanted nothing to do with the rapidly growing comercial UNIX market place. Bob Marsh, John Bass, and several dozen other UNIX supporters met in Santa Clara that year and formed /usr/group which was friendly to suits and capitalists wanting to make a living from UNIX products.

Onyx systems found itself scrambling to rebuild it's UNIX team and product line with the cost reduced C-5002 system. They never recovered, and closed their doors a few years later unable to convince the market they were REALLY serious about UNIX. The linear addressing model of the Motorola 68000 product line sealed their fate since almost every other new UNIX company choose that architecture.


Copyright 1983-1999 by John L. Bass, DMS Design - All rights reserved.